Data processing method and device, storage medium and electronic equipment

Abstract

The invention discloses a data processing method and device, a storage medium and electronic equipment. The method comprises the steps of obtaining historical input data and historical effect data associated with a target application program; constructing an initial channel combination model according to the historical input data and the historical effect data, wherein the initial channel combination model comprises a first transformation sub-model and a second transformation sub-model; performing quasi-experimental analysis on the initial channel combination model to determine an evaluation result of the marketing efficiency of each channel, and evaluating the channel marketing efficiency; and under the condition that the evaluation result shows that the marketing efficiency of each channel meets the preset condition, determining the initial channel combination model as a target channel combination model. The technical problem that the prediction accuracy of the model is low due to the fact that the model for predicting the marketing effect of the marketing channel of the application program is too single is solved, and the embodiment of the invention can be applied to various scenes such as the cloud technology, artificial intelligence, intelligent traffic and auxiliary driving.

Description

Technical Field

[0001] This application relates to the field of computers, and more specifically, to a data processing method and apparatus, a storage medium, and an electronic device. Background Technology

[0002] Currently, related technologies mainly rely on single-channel effect prediction models to predict the marketing effectiveness of channels. Based on the predicted channel performance, data is then strategically allocated to each channel to achieve the desired marketing results. In other words, the channel effect prediction models used in these technologies cannot consider the marketing impact between different channels. They rely solely on a single channel effect prediction model to independently analyze and predict the marketing effectiveness of each channel. As a result, the accuracy of channel prediction marketing is low due to the overly simplistic channel model.

[0003] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0004] This application provides a data processing method and apparatus, storage medium and electronic device to at least solve the technical problem that the prediction accuracy of the model is low due to the overly simplistic model used to predict the marketing effectiveness of the application's marketing channels.

[0005] According to one aspect of the embodiments of this application, a data processing method is provided, comprising: acquiring historical input data and historical effect data associated with a target application, wherein the target application is configured to conduct marketing activities through a set of channels, the historical input data represents the marketing input allocated to the set of channels, and the historical effect data represents the marketing effect generated by the set of channels conducting marketing activities according to the historical input data; constructing an initial channel combination model based on the historical input data and the historical effect data, wherein the initial channel combination model includes a first transformation sub-model and a second transformation sub-model, the first transformation sub-model being used to determine the cumulative impact of marketing activities conducted by each channel on the channel marketing effect in a future period, and the second transformation sub-model being used to determine the degree of reduction in the increase in channel marketing effect as the channel marketing input increases; performing a quasi-experimental analysis on the initial channel combination model to determine the evaluation result of the initial channel combination model for the marketing efficiency of each channel, wherein the quasi-experimental analysis includes determining the predicted marketing effect of each channel based on the historical input data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and the historical effect data; and determining the initial channel combination model as a target channel combination model when the evaluation result indicates that the marketing efficiency of each channel meets preset conditions.

[0006] According to another aspect of the embodiments of this application, a data processing apparatus is also provided, comprising:

[0007] The acquisition module is used to acquire historical investment data and historical performance data associated with the target application, wherein the target application is configured to conduct marketing campaigns through a set of channels, the historical investment data represents the marketing investment allocated to the set of channels, and the historical performance data represents the marketing performance generated by the set of channels conducting marketing campaigns according to the historical investment data.

[0008] The module is used to build an initial channel combination model based on historical investment data and historical performance data. The initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities carried out in each channel on the channel marketing performance in future periods. The second transformation sub-model is used to determine the degree to which the increase in channel marketing performance decreases as channel marketing investment increases.

[0009] The analysis module is used to conduct quasi-experimental analysis on the initial channel mix model to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel. The quasi-experimental analysis includes determining the predicted marketing effect of each channel based on historical investment data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and historical effect data.

[0010] The determination module is used to determine the initial channel combination model as the target channel combination model when the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions.

[0011] Optionally, the device is used to construct an initial channel combination model based on historical input data and historical effect data in the following manner: performing feature extraction on the historical input data to determine the explanatory variables and control variables corresponding to the initial channel combination model, wherein the explanatory variables represent the independent variables of the initial channel combination model, and the control variables represent the constants of the initial channel combination model; performing Bayesian estimation on the model parameters of the initial channel combination model based on the explanatory variables and control variables to determine the initial channel combination model.

[0012] Optionally, the device is used to estimate the model parameters of the initial channel combination model based on explanatory and control variables in the following manner to determine the initial channel combination model: obtaining the variable types of explanatory and control variables; presetting the initial model parameters according to the variable types, wherein, when the variable type includes media variables, the prior distribution corresponding to the initial model parameters is a non-negative distribution; when the variable type includes the decay rate of channel marketing effect of different channels, the prior distribution corresponding to the initial model parameters is determined by the decay rate; when the variable type includes the peak value of channel marketing effect of different channels, the prior distribution corresponding to the initial model parameters is determined by the peak value; determining the target likelihood function based on historical input data and historical effect data, wherein the target likelihood function is used to determine the probability distribution of historical effect data obtained based on historical input data when the initial model parameters are set; updating the target model parameters according to the target likelihood function and the initial model parameters to determine the initial channel combination model, wherein the target model parameters represent the posterior distribution determined according to the target likelihood function and the initial model parameters.

[0013] Optionally, the device is used to perform quasi-experimental analysis on the initial channel combination model in the following manner to determine the evaluation results of the marketing efficiency of the initial channel combination model for each channel: determining a first set of media features and a second set of media features corresponding to each channel, wherein the first set of media features represents media features that belong to explanatory variables and control variables, and the second set of media features represents media features that do not belong to explanatory variables and control variables, with one media feature corresponding to one channel; inputting the channel marketing investment corresponding to the first set of media features from historical investment data into a first transformation sub-model to determine the first sub-transformation parameter corresponding to each channel corresponding to the first set of media features; inputting the first sub-transformation parameter into a second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel; multiplying the second sub-transformation parameter corresponding to each channel with the channel elasticity parameter corresponding to each channel to determine the channel transformation value of the channel marketing investment corresponding to each channel, wherein the channel elasticity parameter represents the magnitude of change in channel marketing effect with changes in channel marketing investment, and the initial channel combination model obtains the target channel combination model by adjusting the channel elasticity parameter; determining the predicted marketing effect by summing the channel transformation values ​​of the channel marketing investment corresponding to each channel; and determining the evaluation result based on a comparison between the predicted marketing effect and historical effect data.

[0014] Optionally, the device is used to input the channel marketing investment corresponding to the first group of media characteristics from the historical investment data into the first transformation sub-model in the following manner to determine the first sub-transformation parameter corresponding to each channel: obtaining the channel marketing investment corresponding to the first group of media characteristics for L unit statistical periods for each channel, where L represents the total duration of the marketing influence of a channel; obtaining the attenuation parameter and the peak influence parameter pre-set for each channel, where the attenuation parameter represents the rate of attenuation of the channel marketing effect corresponding to a channel over time, and the peak influence parameter represents the lag time relative to the initial placement when the channel marketing effect corresponding to a channel reaches its peak; and determining the first sub-transformation parameter corresponding to each channel based on the channel marketing investment, attenuation parameter, and peak influence parameter corresponding to the first group of media characteristics.

[0015] Optionally, the device is used to input the first sub-variation parameter corresponding to each channel into the second transformation sub-model in the following manner to determine the second sub-transformation parameter corresponding to each channel: when the channel marketing effects between each channel do not affect each other, the first sub-variation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using a saturated transformation function; when the channel marketing effects between each channel affect each other, the first sub-variation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using a target hyperparameter determined by the channel characteristics.

[0016] Optionally, the device is used to determine the evaluation result by comparing predicted marketing effect and historical effect data in the following manner: determining the predicted marketing effect for each channel based on the predicted marketing effect, channel marketing investment, and channel elasticity parameters; determining the actual channel marketing effect for each channel based on historical investment data and historical effect data; and determining the evaluation result for each channel based on the magnitude relationship between the predicted marketing effect and the actual channel marketing effect.

[0017] Optionally, the device is used to determine the actual channel marketing effect corresponding to each channel based on historical input data and historical effect data in the following manner: determining first historical input data and first historical effect data for which the target channel did not participate in the campaign, and second historical input data and second historical effect data for which the target channel participated in the campaign, wherein the target channel represents any one of the channels; training an initial marketing effect prediction model based on the first historical input data and first historical effect data to obtain a target marketing effect prediction model; inputting the second historical input data into the target marketing effect prediction model to determine the counterfactual predicted marketing effect corresponding to the target channel; and determining the actual channel marketing effect of the target channel based on the second historical effect data and the counterfactual predicted marketing effect.

[0018] Optionally, the device is used to determine the predicted marketing effect for each channel based on the predicted marketing effect, channel marketing investment, and channel elasticity parameters in the following manner: determining the target channel marketing investment and target channel elasticity parameters corresponding to the target channel based on the channel marketing investment and channel elasticity parameters; determining a first intermediate value with the target channel marketing investment as the base and the target channel elasticity parameters as the exponent; determining a second intermediate value as the ratio of the predicted marketing effect to the first intermediate value; determining the target predicted marketing effect of the target channel as the difference between the predicted marketing effect and the second intermediate value; the device is used to determine the evaluation result for each channel based on the relationship between the predicted marketing effect and the actual channel marketing effect in the following manner: when the actual channel marketing effect is less than or equal to the target predicted marketing effect, determining the target evaluation result for the target channel indicates that the marketing efficiency of the target channel meets the preset conditions; when the actual channel marketing effect is greater than the target predicted effect, determining the target evaluation result indicates that the marketing efficiency of the target channel does not meet the preset conditions.

[0019] Optionally, the device is used to perform feature extraction operations on historical input data to determine the explanatory and control variables corresponding to the initial channel combination model by: determining the correlation coefficient between the channel marketing input and historical performance data of each channel in a set of channels based on the historical input data, wherein the correlation coefficient is used to represent the correlation between each channel and the historical performance data; performing feature extraction operations on channels whose correlation coefficients meet preset sorting conditions to determine the explanatory and control variables corresponding to the initial channel combination model.

[0020] Optionally, the apparatus is further configured to: after determining the initial channel combination model as the target channel combination model, obtain first expected input data, wherein the first expected input data includes the total amount of expected input to a group of channels; input the first expected input data into the target channel combination model, and allocate the first expected input data based on the marketing efficiency of each channel in the group of channels through the target channel combination model to generate first expected effect data, wherein the first expected effect data represents the maximum marketing effect that can be obtained by marketing according to the first expected input data.

[0021] Optionally, the apparatus is further configured to: after determining the initial channel combination model as the target channel combination model, obtain second expected effect data, wherein the second expected effect data includes the total expected effect obtained from a group of channels; input the second expected effect data into the target channel combination model, and determine the second expected input data allocated to the group of channels based on the marketing efficiency of each channel in the group of channels through the target channel combination model, wherein the second expected input data represents the minimum marketing input required to obtain the second expected effect data.

[0022] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, wherein a computer program is stored in the computer-readable storage medium, and the computer program is configured to execute the above-described data processing method when running.

[0023] According to another aspect of the embodiments of this application, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the data processing method described above.

[0024] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to execute the data processing method described above through the computer program.

[0025] In this embodiment, historical input data and historical effect data associated with the target application are acquired. The target application is configured to conduct marketing activities through a set of channels. The historical input data represents the marketing input allocated to the set of channels, and the historical effect data represents the marketing effect generated by the marketing activities conducted by the set of channels according to the historical input data. An initial channel combination model is constructed based on the historical input data and historical effect data. The initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities conducted on each channel on the channel marketing effect in future periods. The second transformation sub-model is used to determine the degree to which the increase in channel marketing effect decreases as channel marketing input increases. A quasi-experimental analysis is performed on the initial channel combination model to determine the evaluation results of the initial channel combination model for the marketing efficiency of each channel. This involves determining the predicted marketing effectiveness of each channel based on historical input data, and evaluating the marketing efficiency of each channel based on the predicted marketing effectiveness and historical performance data. If the evaluation results indicate that the marketing efficiency of each channel meets preset conditions, the initial channel combination model is determined as the target channel combination model. An initial channel combination model, including a first transformation sub-model and a second transformation sub-model, is constructed using historical input and performance data. Furthermore, a quasi-experimental analysis is conducted on this initial channel combination model to evaluate the marketing effectiveness and efficiency of each channel. Finally, the target channel combination model is obtained based on the evaluation results. This achieves the technical effect of predicting the marketing effectiveness of each channel using the target channel combination model, thereby improving the marketing efficiency of each channel. This also solves the technical problem of low prediction accuracy caused by overly simplistic models for predicting the marketing effectiveness of application marketing channels.

[0026] On the other hand, in this application, the first transformation sub-model in the initial channel mix model can determine the cumulative impact of channel marketing effects in future periods based on the marketing activities carried out by each channel. Furthermore, through the second transformation sub-model in the initial channel mix model, the relationship between the marketing activities carried out by each channel and the channel marketing investment can be obtained. In turn, the correlation between the channel marketing effect and the marketing activities and channel marketing investment can be determined, so as to determine the interaction between different channels, improve the prediction accuracy of the target evaluation channel model, and achieve the purpose of optimizing the channel placement strategy. Attached Figure Description

[0027] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0028] Figure 1 This is a schematic diagram of an application environment for an optional data processing method according to an embodiment of this application;

[0029] Figure 2 This is a flowchart illustrating an optional data processing method according to an embodiment of this application;

[0030] Figure 3 This is a schematic diagram of an optional data processing method according to an embodiment of this application;

[0031] Figure 4 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0032] Figure 5 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0033] Figure 6 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0034] Figure 7 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0035] Figure 8 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0036] Figure 9 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0037] Figure 10 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0038] Figure 11 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0039] Figure 12 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0040] Figure 13 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0041] Figure 14 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0042] Figure 15 This is a schematic diagram of another optional data processing method according to an embodiment of this application;

[0043] Figure 16 This is a schematic diagram of the structure of an optional data processing apparatus according to an embodiment of this application;

[0044] Figure 17 This is a schematic diagram of the structure of an optional data processing product according to an embodiment of this application;

[0045] Figure 18 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0046] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0047] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0048] The present application will be described below with reference to embodiments:

[0049] According to one aspect of the embodiments of this application, a data processing method is provided. Optionally, in this embodiment, the above-described data processing method can be applied to, for example... Figure 1 The hardware environment shown consists of server 101 and terminal device 103. For example... Figure 1As shown, server 101 is connected to terminal device 103 via a network and can be used to provide services to terminal device or applications installed on terminal device. Application 107 can be video application, instant messaging application, browser application, educational application, game application, etc. Database 105 can be set up on the server or independently of the server to provide data storage services for server 101, such as a game data storage server. The network mentioned above can include, but is not limited to, wired networks and wireless networks. The wired network includes local area networks, metropolitan area networks, and wide area networks. The wireless network includes Bluetooth, WIFI, and other networks that enable wireless communication. Terminal device 103 can be a terminal configured with an application, and can include, but is not limited to, at least one of the following: mobile phones (such as Android phones, iOS phones, etc.), laptops, tablets, handheld computers, MID (Mobile Internet Devices), PADs, desktop computers, smart TVs, smart voice interaction devices, smart home appliances, vehicle terminals, aircraft, virtual reality (VR) terminals, augmented reality (AR) terminals, mixed reality (MR) terminals, and other computer devices. The server mentioned above can be a single server, a server cluster composed of multiple servers, or a cloud server.

[0050] Combination Figure 1 As shown, the above data processing method can be executed by an electronic device, which can be a terminal device or a server. The above data processing method can be implemented by the terminal device or the server respectively, or by the terminal device and the server together.

[0051] The above is merely an example, and this embodiment does not impose any specific limitations.

[0052] Alternatively, as an alternative implementation method, such as Figure 2 As shown, the data processing methods include:

[0053] S202, Obtain historical investment data and historical performance data associated with the target application, wherein the target application is configured to conduct marketing campaigns through a set of channels, the historical investment data represents the marketing investment allocated to the set of channels, and the historical performance data represents the marketing performance generated by the set of channels conducting marketing campaigns according to the historical investment data;

[0054] Optionally, in this embodiment, the target application can include, but is not limited to, systems, platforms, etc., and can be used to manage and execute marketing activities, including but not limited to email marketing tools, social media management platforms, advertising delivery systems, etc. Historical investment data represents the funds and resources allocated to various channels in marketing activities. This can help analyze historical marketing strategies and effects, thereby providing a basis for future decision-making. Historical investment data includes, but is not limited to: advertising expenditure: advertising costs invested on specific channels; time investment: time and human resources spent on specific channels; material investment: costs for producing and distributing marketing materials; social media activities: promotional activities and interactions conducted on social media platforms; email marketing: the number and type of emails sent to the target audience; historical performance data: this refers to the actual effect of a marketing campaign conducted on a group of channels based on historical investment data; historical investment data can be used to evaluate the success of marketing activities and provide a reference for optimizing future marketing strategies; historical performance data includes, but is not limited to: click-through rate (CTR): the number of times a user clicks on an advertisement or marketing content. The percentage of users who click on an ad or marketing content and then complete a target action (such as purchasing or registering) is categorized as follows: Conversion rate: The percentage of users who complete a target action (such as purchasing or registering) after clicking on the ad or marketing content; Social media engagement: The number of likes, comments, and shares on social media platforms; Email open rate: The percentage of recipients who open an email; Customer acquisition cost (CAC): The average cost required to acquire a new customer; Channels: In marketing campaigns, channels refer to the platforms or tools used to disseminate information and interact with the target audience. Channels can include, but are not limited to, traditional media: such as television, radio, and newspapers; digital media: such as search engines, social media, and email marketing; mobile applications: such as in-app push notifications and advertisements; and offline events: such as exhibitions, conferences, and promotional activities.

[0055] For example, Figure 3 This is a schematic diagram of an optional data processing method according to an embodiment of this application. The aforementioned historical input data and historical effect data may include, but are not limited to, the following: Figure 3 As shown, the characteristics are divided into media-related characteristics, operational characteristics, industry characteristics, and natural characteristics.

[0056] S204. Construct an initial channel mix model based on historical investment data and historical performance data. The initial channel mix model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities carried out in each channel on the channel marketing performance in future periods. The second transformation sub-model is used to determine the degree to which the increase in channel marketing performance decreases as channel marketing investment increases.

[0057] Optionally, in this embodiment, the aforementioned channel marketing effectiveness may include, but is not limited to, various indicators such as sales revenue, market share, customer satisfaction, brand awareness, and conversion rate. Different enterprise users can choose different effectiveness indicators according to their business objectives and measurement standards. The impact refers to the total effect of a marketing activity throughout the entire marketing cycle, not just the short-term impact. Channel marketing investment may include, but is not limited to, the investment of various resources, such as advertising expenses, human resource costs, and time costs. The first transformation sub-model and the second transformation sub-model can be mathematical or statistical models used to transform input data (such as marketing investment) into output data (such as marketing effectiveness). The first transformation sub-model and the second transformation sub-model can be used to predict the relationship between different inputs and outputs of the model.

[0058] For example, historical marketing campaign data is first collected and analyzed, including but not limited to the resources invested (such as advertising expenditure, manpower, etc.) and the final marketing results (such as sales, customer satisfaction, etc.) to construct an initial channel mix model. This initial channel mix model consists of a first transformation sub-model and a second transformation sub-model, each of which has its specific function.

[0059] Specifically, the first transformation sub-model can assess the impact of different marketing channel activities on the entire marketing cycle. For example, if an advertising campaign is conducted on website A, the first transformation sub-model can determine the impact of the advertising campaign on the marketing effectiveness of website A throughout the entire marketing cycle. The second transformation sub-model will assess how the rate of increase in marketing effectiveness of a channel slows down as the amount of channel resources invested. In other words, as investment increases, the increase in effectiveness brought about by each additional unit of investment gradually decreases.

[0060] For example, the first transformation sub-model mentioned above may include, but is not limited to, the Adstock model. The first transformation sub-model is used to measure the cumulative impact (cumulative impact parameter) of advertising on a certain channel on the future sales or brand awareness of a certain product. For example, assuming that advertising not only affects the conversion rate in the current period, but also has an impact on several future periods, and that this impact gradually weakens over time, the first transformation sub-model may include, but is not limited to, at least the following two forms: one is an exponentially decreasing form, and the other is a decreasing form with a peak:

[0061] Adstock1 represents an exponentially decreasing form with no peak value, calculated as follows:

[0062] A t =Adstock(x t ,…,x t-(L-1) ;L,D)

[0063] =X t +D*Xt-1 +D 2 *X t-2 +…+D L *X t-1

[0064] Among them: A t X represents the Adstock value in period t. t-i Let represent the amount of advertising spent in period ti, D represent the decay coefficient (0≤D≤1), L represent the rate at which the advertising effect decays over time, and L represent the maximum time delay of the advertising effect.

[0065] Adstock2 indicates a decreasing form with a peak, calculated as follows:

[0066]

[0067]

[0068] Among them: A t X represents the Adstock value in period t. t-i Let represent the amount of advertising placed in period ti, D represent the decay coefficient (0≤D≤1), which represents the rate at which the advertising effect decays over time, L represent the maximum time delay of the advertising effect, and P represent the peak / delay of the media effect, which is the time period lagging behind the first exposure.

[0069] Adstock3 also represents a decreasing form with a peak, calculated as follows:

[0070]

[0071] Among them: A t X represents the Adstock value in period t. t-i Let t represent the input data for period ti, D represent the decay coefficient (0≤D≤1), which represents the rate at which the advertising effect decays over time, L represent the maximum time delay of the advertising effect, and P represent the peak / delay of the media effect, which is the time period lagging behind the first exposure.

[0072] Further, Figure 4 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 4 As shown, experiments can be conducted on the changes of Adstock1, Adstock2, and Adstock3 mentioned above, and the final calculated A values ​​for different transformation forms can be obtained. t There are also differences. A reasonable range of hyperparameters can be set based on the specific business scenario and different channels. For example, Adstock2 and Adstock3 are more suitable for this application scenario.

[0073] For example, the second transformation sub-model mentioned above may include, but is not limited to, the Saturation Transformation model, where Saturation Transformation can be used to describe the phenomenon that the advertising effect changes as investment increases. That is, after advertising expenditure reaches a certain level, its incremental effect on sales will gradually weaken and eventually reach a saturation point.

[0074] For example, Figure 5 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 5 As shown, the second transformation sub-model can be represented as the Hill function:

[0075]

[0076] Where: x A,standard γ represents the marketing investment of channel A after standardization. hill The value represents the theoretical potential user base of the channel. K represents the X value corresponding to the maximum marginal benefit of the channel. The smaller K is, the earlier the inflection point appears. S represents the steepness of the curve, which determines the speed at which the curve reaches the saturation point. The higher the S value, the easier it is to reach the saturation point. Insights and understandings about consumer behavior, market trends, and competitors obtained through data analysis, market research, customer feedback, etc., help companies make more informed decisions and formulate more effective strategies. It can be determined that channels with high S values ​​are suitable for short-term investment, while channels with low S values ​​are suitable for long-term and continuous investment.

[0077] also, Figure 6 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 6 As shown, taking the second transformation sub-model as a multiplicative model as an example, it can also be determined using the log function and the standardized parameter scalar, making the second transformation sub-model more consistent with actual business applications. When marketing investment is small, there will be no excessive marginal effect. In this case, the second sub-transformation model can be expressed as:

[0078] X′ A =log(x A +C), where x A This represents the input data for channel A, and C represents the scalar constant, a hyperparameter that is mainly determined by channel characteristics (e.g., the potential user base of the channel, the reach efficiency of the channel, etc.).

[0079] S206, Conduct a quasi-experimental analysis on the initial channel mix model to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel. The quasi-experimental analysis includes determining the predicted marketing effect of each channel based on historical investment data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and historical effect data.

[0080] Optionally, in the embodiments of this application, quasi-experimental analysis may include, but is not limited to, being used to assess causal relationships, especially when a randomized controlled experiment cannot be conducted. Quasi-experimental analysis attempts to control other variables through statistical methods in order to more accurately assess the treatment effect; historical input data may include, but is not limited to, investment and resource allocation data on various marketing channels, such as advertising expenditure, number of content releases, and campaign scale; predicting marketing effects may include, but is not limited to, using statistical models or machine learning algorithms based on historical input data to predict the future marketing effects of various channels, such as predicting customer conversion rates, sales, and brand awareness; historical effect data may include, but is not limited to, actual effect data of historical marketing activities, such as customer conversion rates, sales, and brand awareness, which can be used to evaluate and verify the accuracy of the predictive model; marketing efficiency evaluation may include, but is not limited to, evaluating the marketing efficiency of various channels based on predicted marketing effects and historical effect data, and may include, but is not limited to, comparing indicators such as return on investment (ROI), cost-benefit analysis (CBA), or net present value (NPV) of different channels.

[0081] For example, quasi-experimental analysis can be used to determine channel efficiency and calibrate the initial channel combination model. Taking channel A as an example, historical investment data can identify periods when data was continuously invested in channel A, periods with unusual investments (such as a sudden increase in investment), and periods when no data was invested in channel A. In quasi-experimental analysis, channels are first screened out by anomaly detection, corresponding samples are constructed for each channel, and model training and counterfactual inference are performed to finally obtain the channel efficiency range. Incremental experiments can also be performed on channels to obtain more accurate channel efficiency data.

[0082] In one exemplary embodiment, it should be noted that each of the above-mentioned channels can be used for quasi-experimental analysis, including but not limited to:

[0083] S1, obtain historical input data and historical effect data for a period of time;

[0084] S2, find the first historical performance data where channel A did not participate and the second historical performance data where channel A participated from the historical performance data;

[0085] S3, the model is trained based on the first historical effect data, so that the model can fit the effect data according to the input data;

[0086] S4. Input the second historical input data corresponding to the second historical effect data into the above model to obtain the predicted effect data. The predicted effect data represents the predicted result of the effect data in which channel A did not participate.

[0087] S5, determine the actual value based on the second historical effect data and the predicted effect data;

[0088] S6. Compare the actual value with the predicted value of the initial channel combination model. If the predicted value of the initial channel combination model is greater than the actual value, it means that the initial channel combination model can be determined as the target channel combination model.

[0089] For example, the initial channel combination model described above may include, but is not limited to:

[0090]

[0091] Where, organic factor represents the control variable, Y total The model represents the relationship between the overall new arrivals and resources invested in the application, and is used to solve the overall resource allocation. α can be set flexibly, and this application does not impose any restrictions on it.

[0092] The above When presented in an additive model, it can be represented as:

[0093]

[0094] Where i represents the media variable among the explanatory variables, j represents the media variable among the control variables, and the classification of media variables and non-media variables is as follows: Figure 3 As shown, α represents the intercept term, which can be understood as the intercept term when y t When y = α, t This represents the basic amount of effect data for a time period t when no data was input; β i Let L, P, and D be the elasticity coefficients of the channel corresponding to media variable i; L, P, and D are the Adstock transformation parameters; and S represents the transformation parameters of the second sub-transformation model.

[0095] It should be noted that the target predicted marketing efficiency for channel A is effect. A , represents the target predicted marketing efficiency of channel A obtained through the initial channel combination model, and S represents the transformation parameters of the second sub-transformation model.

[0096] effect A =β A ×Saturation(Adstock(Scale(X A );L A ,PA D A ),S A )

[0097] Among them, L A ,P A D A S represents the Adstock transformation parameter of channel A. A The saturation transformation parameter of channel A is represented.

[0098] The above When presented in a multiplication model, it can be represented as:

[0099]

[0100] Among them, X i,t Let logα represent the marketing investment of the channel corresponding to media variable i in period t, and β represent the intercept term (which can be 0). i This represents the elasticity coefficient of the channel corresponding to media variable i, that is, the impact of changing the investment in channel i by 1% at the current level on β. i The percentage change of y, L, P, D are Adstock transformation parameters, and C represents the scaling constant (hyperparameter), which can be flexibly set, for example, determined by channel characteristics (such as the potential user base of the channel, reach efficiency, etc.).

[0101] It should be noted that the target predicted marketing efficiency for channel A is effect. A , represents the target predicted marketing efficiency of channel A obtained through the initial channel combination model, and S represents the transformation parameters of the second sub-transformation model.

[0102] Understandably, the model in the initial channel combination model above can be fitted to obtain:

[0103]

[0104] S208, if the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the initial channel combination model is determined as the target channel combination model.

[0105] Optionally, in this embodiment, the evaluation result represents the analysis and evaluation of the efficiency of the marketing channels, which may include, but is not limited to, the statistics and comparison of various data to determine the performance of each channel; marketing efficiency can be understood as the ability of a marketing channel to achieve the expected marketing goals, usually measured by indicators such as return on investment and conversion rate; preset conditions represent the goals or expectations set by the enterprise and users in advance before starting the marketing activity, such as expected sales volume, customer acquisition cost, etc.; the target channel combination model means that after actual operation and evaluation, if each channel of the initial channel combination model has achieved the preset efficiency conditions, then the initial channel combination model will be identified as the target channel combination model, that is, the enterprise and users will carry out marketing activities according to the target channel combination model.

[0106] Specifically, the process of evaluating channel marketing efficiency may include, but is not limited to, steps such as model training and prediction, and effectiveness quantification (Lift value calculation):

[0107] First, a prediction model (such as RF, XGB, Lasso, etc.) is trained using data from the control group to construct a prediction model M based on normal delivery conditions. Then, the data from the experimental group (T) is applied to this prediction model to obtain the counterfactual prediction value y. pred Secondly, by comparing the observed values ​​y in the experimental group... observed The counterfactual outcome value y predicted by the model pred Calculate the channel lift value. The formula for calculating the lift value is: Lift = y observed -y pred The purpose is to evaluate the actual marketing effectiveness of each channel. If the actual marketing efficiency of each channel meets the preset conditions (e.g., reaches a certain lift value), then the initial channel combination model can be considered effective.

[0108] Specifically, an effective initial channel mix model can be identified as the target channel mix model, and its values ​​can be used to guide and optimize channel placement strategies. That is, after a series of evaluation processes (including model training, prediction, and effect quantification), if the marketing efficiency of each channel meets the preset conditions, the initial channel mix model can be used as the target channel mix model to facilitate future marketing decisions and optimization.

[0109] For example, Figure 7 This is a schematic diagram of another optional data processing method according to an embodiment of this application, which obtains the effect through an initial channel combination model. A Compare the effects AIf the predicted value of the initial channel combination model is greater than the true value mentioned above, it indicates that the initial channel combination model is usable and can be identified as the target channel combination model. This target channel combination model can then be used to generate models such as... Figure 7 The channel analysis diagram shown includes analysis of new entrants, channel cost-effectiveness analysis, content marketing, channel efficiency and marginal benefits, the flow relationship between advertising investment and acquisition channels, ecosystem effects, how to allocate resources to maximize expected investment data for the next cycle given the investment resources for the next cycle; and how to allocate resources to maximize customer acquisition for the next cycle given the expected investment data for the next cycle.

[0110] Specifically, Figure 8 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 8 As shown, this illustrates the contribution breakdown and trend description of key business metrics data for the application within the current period.

[0111] For example, Figure 9 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 9 As shown, the channel cost-effectiveness analysis scenario can be intuitively displayed on the terminal device's display interface through the embodiments of this application.

[0112] For example, Figure 10 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 10 As shown, the efficiency of content marketing in improving user acquisition, organic traffic, and overall traffic can be intuitively demonstrated on the display interface of the terminal device through the embodiments of this application.

[0113] For example, Figure 11 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 11 As shown, channel efficiency and marginal benefits can be intuitively viewed on the display interface of the terminal device through the embodiments of this application.

[0114] For example, Figure 12 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 12 As shown, the flow relationship between advertising investment and revenue channels can be intuitively viewed on the display interface of the terminal device through the embodiments of this application.

[0115] For example, Figure 13 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 13As shown, the ecological effect of the channel can be intuitively viewed on the display interface of the terminal device through the embodiments of this application, that is, the marketing effect that the channel can obtain without the input of data.

[0116] For example, Figure 14 This is a schematic diagram of another optional data processing method according to an embodiment of this application, such as... Figure 14 As shown, the terminal device's display interface allows for a direct view of the resources invested in the next cycle, how to allocate resources to maximize the expected investment data for the next cycle, and how to allocate resources to maximize customer acquisition for the next cycle, given the expected investment data for the next cycle.

[0117] For example, Figure 15 This is a schematic diagram of another optional data processing method according to an embodiment of this application. In game applications, the data processing method proposed in this application can be used to determine the marketing strategy of the game application, including but not limited to... Figure 15 The initial channel mix model is established (1502), and the initial channel mix model is adjusted through experiment (1504) to obtain the target channel mix model. Furthermore, the target channel mix model is used for marketing strategy decision-making and analysis (1506).

[0118] This application's embodiments involve acquiring historical investment data and historical performance data associated with a target application. The target application is configured to conduct marketing activities through a set of channels. Historical investment data represents the marketing investment allocated to the set of channels, and historical performance data represents the marketing performance generated by the marketing activities conducted by the set of channels according to the historical investment data. An initial channel combination model is constructed based on the historical investment data and historical performance data. This initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities conducted through each channel on the channel marketing performance in future periods. The second transformation sub-model is used to determine the degree to which the increase in channel marketing performance decreases as channel marketing investment increases. A quasi-experimental analysis is performed on the initial channel combination model to determine the evaluation results of the initial channel combination model's marketing efficiency for each channel. The quasi-experimental analysis includes determining the predicted marketing effect of each channel based on historical input data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and historical effect data. If the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the initial channel combination model is determined as the target channel combination model. An initial channel combination model, including a first transformation sub-model and a second transformation sub-model, is constructed using historical input and historical effect data. Furthermore, a quasi-experimental analysis is conducted on this initial channel combination model to evaluate the marketing effect and efficiency of each channel. Finally, the target channel combination model is obtained based on the evaluation results. This achieves the technical effect of using the target channel combination model to predict the marketing effect of each channel, improving the marketing efficiency of each channel, and thus solving the technical problem of low predictive marketing accuracy due to overly simplistic channel models.

[0119] On the other hand, in this application, the first transformation sub-model in the initial channel mix model can determine the cumulative impact of channel marketing effects in future periods based on the marketing activities carried out by each channel. Furthermore, through the second transformation sub-model in the initial channel mix model, the relationship between the marketing activities carried out by each channel and the channel marketing investment can be obtained. In turn, the correlation between the channel marketing effect and the marketing activities and channel marketing investment can be determined, so as to determine the interaction between different channels, improve the prediction accuracy of the target evaluation channel model, and achieve the purpose of optimizing the channel placement strategy.

[0120] As an optional approach, an initial channel combination model is constructed based on historical input data and historical effect data. This includes: performing feature extraction on the historical input data to determine the explanatory and control variables corresponding to the initial channel combination model, where the explanatory variables represent the independent variables of the initial channel combination model and the control variables represent the constants of the initial channel combination model; and performing Bayesian estimation on the model parameters of the initial channel combination model based on the explanatory and control variables to determine the initial channel combination model.

[0121] Optionally, in this embodiment, the Bayesian estimation described above is a statistical method that considers the prior distribution of the parameters during parameter estimation. Bayesian estimation considers not only observed data but also prior knowledge of the parameters. By calculating the posterior distribution, Bayesian estimation can update the estimated values ​​of the parameters. The basic idea of ​​Bayesian estimation is to use Bayes' theorem to combine the prior distribution with the likelihood function to obtain the posterior distribution of the parameters:

[0122]

[0123] Where P(β|Y) is the posterior distribution of parameter β, P(Y|β) is the likelihood function, P(β) is the prior distribution, and P(Y) is the probability of the observed data.

[0124] For example, when performing feature extraction on historical input data, it is necessary to determine the explanatory and control variables of the initial channel mix model. Explanatory variables are the independent variables of the model, which affect the target variable; control variables are constants in the model, remaining unchanged throughout. Next, the model parameters are estimated using OLS or Bayesian methods to determine the initial channel mix model.

[0125] It should be noted that the model parameters of the initial channel combination model can be determined through Bayesian estimation. These parameters include, but are not limited to, the intercept term alpha, the regression coefficients of the variables, the Adstock transformation parameters, and the saturation transformation parameters. This includes, but is not limited to, estimation using the Hierarchical Bayesian method.

[0126] First, determine the prior distribution of the model parameters. This represents the general understanding of the parameters before observing data (e.g., the parameters follow a normal distribution with a mean of 0 and a standard deviation of 1). First, determine the form of the prior distribution. For example, for the regression coefficients of media variables, the prior distribution can be set to Half-Normal or other non-negative distributions. The parameter range can also be set based on the characteristics of the channel. Historical data can be used to set an appropriate prior distribution. After observing data and calculating the corresponding likelihood function, the probability of observing the current data can be determined. At this point, the understanding of the model parameters can be updated based on the observed situation and the likelihood function, i.e., the calculated posterior distribution. In Bayesian estimation, strong and weak prior distributions have a significant impact on parameter estimation. Taking HillSaturation parameter transformation and Adstock parameter transformation as examples, an overly broad prior distribution will result in slower parameter convergence, while an overly strict prior distribution may cause the posterior distribution to deviate too far from the true value. The selection of the prior distribution can be categorized according to its usefulness, including but not limited to the following:

[0127] FlatPrior: Suitable for use when there is no information about the parameters and will not affect the parameter estimation.

[0128] Super-vaguebutproperty applies to cases where the prior distribution has certain mathematical properties (e.g., integrability, which means that the product of the prior distribution and the likelihood function can be integrated to obtain the posterior distribution).

[0129] Weeklyinformativeprior is suitable for situations where there are rough estimates of model parameters that do not need to affect the posterior distribution;

[0130] When there is a strong influence on the posterior distribution, Generic weekly information can be selected;

[0131] If you need to introduce some general information, you can choose Specificinformationprior.

[0132] As an optional approach, the model parameters of the initial channel combination model are estimated based on explanatory and control variables to determine the initial channel combination model. This includes: obtaining the variable types of explanatory and control variables; pre-setting the initial model parameters according to the variable types, wherein, when the variable type includes media variables, the prior distribution of the initial model parameters is a non-negative distribution; when the variable type includes the decay rate of channel marketing effectiveness of different channels, the prior distribution of the initial model parameters is determined by the decay rate; when the variable type includes the peak value of channel marketing effectiveness of different channels, the prior distribution of the initial model parameters is determined by the peak value; determining the target likelihood function based on historical input data and historical effect data, wherein the target likelihood function is used to determine the probability distribution of historical effect data obtained based on historical input data, given the initial model parameters; updating the target model parameters based on the target likelihood function and the initial model parameters to determine the initial channel combination model, wherein the target model parameters represent the posterior distribution determined based on the target likelihood function and the initial model parameters.

[0133] Optionally, in the embodiments of this application, in statistical modeling, the explanatory variables (also called independent variables or predictor variables) are the variables used in the model to predict the outcome. Control variables are those variables that may affect the outcome but are not the main focus of the study; variable types can be divided into different types according to the nature of the variables, such as... Figure 3 As shown, this variable type can include, but is not limited to, the peak value of the channel marketing effect corresponding to each channel, and then the prior distribution corresponding to the initial model parameters can be determined based on this variable type; when building the model, it is necessary to set initial values ​​for the parameters in the model. The initial values ​​of these parameters will affect the model fitting process and the final result.

[0134] It's important to note that the prior distribution, in Bayesian statistics, can be understood as the researcher's belief distribution regarding the parameters before seeing the data. A nonnegative distribution is a common prior distribution, suitable for parameters that can only take nonnegative values. The decay rate mentioned above indicates that the value of the model parameter will decrease with time or other factors; this rate of decrease is called the decay rate. The target likelihood function is a method in statistics to measure how well a model fits the data; given the model parameters, the target likelihood function is the probability distribution of the model predicting historical performance data. In Bayesian statistics, the posterior distribution is the belief distribution regarding the parameters after seeing the data.

[0135] As an optional approach, a quasi-experimental analysis is conducted on the initial channel mix model to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel. This includes: determining the first set of media characteristics and the second set of media characteristics corresponding to each channel, where the first set of media characteristics represents media characteristics that are explanatory and control variables, and the second set of media characteristics represents media characteristics that are not explanatory and control variables, with one media characteristic corresponding to one channel; inputting the marketing investment of the channels corresponding to the first set of media characteristics from the historical investment data into the first transformation sub-model to determine the first sub-transformation parameters corresponding to each channel corresponding to the first set of media characteristics; and then inputting the marketing investment of each channel corresponding to the first set of media characteristics into the first transformation sub-model. The first sub-variation parameter corresponding to each channel is input into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel. The product of the second sub-transformation parameter corresponding to each channel and the channel elasticity parameter corresponding to each channel is determined as the channel transformation value of the channel marketing investment for each channel. Here, the channel elasticity parameter represents the magnitude of change in channel marketing effect as the channel marketing investment changes. The initial channel combination model is adjusted to obtain the target channel combination model. The sum of the channel transformation values ​​of the channel marketing investment corresponding to each channel is determined as the predicted marketing effect. The evaluation result is determined by comparing the predicted marketing effect with historical effect data.

[0136] For example, the above media features are as follows Figure 3 As shown, when determining the first and second sets of media features corresponding to each channel, the media features need to be classified. Media features that belong to both explanatory and control variables are assigned to the first set, while those that do not belong to either are assigned to the second set. Each media feature corresponds to one channel; that is, each channel has a corresponding set of media features. Next, the marketing investment of the channels corresponding to the first set of media features from historical investment data is input into the first transformation sub-model. The first sub-transformation parameters corresponding to each channel corresponding to the first set of media features are determined through model transformation.

[0137] Furthermore, the first sub-variation parameter is input into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel. The product of the second sub-transformation parameter corresponding to each channel and the channel elasticity parameter corresponding to each channel is the channel transformation value of the channel marketing investment for each channel. The channel elasticity parameter represents the magnitude of change in channel marketing effectiveness as channel marketing investment changes, and is an important parameter for measuring changes in channel marketing effectiveness. The initial channel mix model obtains the target channel mix model by adjusting the channel elasticity parameter, thereby predicting the marketing investment for each channel. The sum of the channel transformation values ​​of the channel marketing investment corresponding to each channel is determined as the predicted marketing effect. Finally, the predicted marketing effect is compared with historical effect data, and the evaluation results are determined through comparative analysis to assess the effectiveness and accuracy of the channel mix model. This process can help companies better understand the marketing effects of different channels, thereby developing more effective marketing strategies.

[0138] As an optional approach, the channel marketing investment corresponding to the first set of media characteristics in the historical investment data is input into the first transformation sub-model to determine the first sub-transformation parameters for each channel. This includes: obtaining the channel marketing investment corresponding to the first set of media characteristics for L statistical periods for each channel, where L represents the total duration of the marketing impact of a channel; obtaining the pre-set decay parameter and peak impact parameter for each channel, where the decay parameter represents the rate at which the channel marketing effect of a channel decays over time, and the peak impact parameter represents the lag time relative to the initial placement when the channel marketing effect of a channel reaches its peak; and determining the first sub-transformation parameters for each channel based on the channel marketing investment, decay parameter, and peak impact parameter corresponding to the first set of media characteristics.

[0139] In an exemplary embodiment, to obtain the channel marketing investment corresponding to the first set of media characteristics for L unit statistical periods for each channel, the following steps may be performed, including but not limited to:

[0140] S1. First, you need to know the initial media characteristics of each channel and the corresponding marketing investment. This data can come from historical data, market research, or expert evaluation.

[0141] S2, L represents the total duration of a marketing impact across a channel. This value can be set according to the actual situation; for example, if the effect of an advertisement lasts for a month, then L might be 30 days.

[0142] S3, the decay parameter (α), represents the rate at which the marketing effectiveness of a channel decays over time. This parameter can be set based on historical data or industry experience.

[0143] S4, Peak Impact Parameter (β), represents the lag time relative to the initial campaign when the channel's marketing effectiveness reaches its peak. This parameter can also be set based on historical data or industry experience.

[0144] S5, based on the channel marketing investment, attenuation parameters, and peak impact parameters corresponding to the first set of media characteristics.

[0145] S6. Once the first sub-transformation parameter is determined, the channel marketing strategy can be adjusted based on this parameter in order to achieve better marketing results.

[0146] S7, after implementing a new marketing strategy, continuously monitors the effectiveness of channel marketing and adjusts the attenuation parameters, peak impact parameters, and first sub-transformation parameters based on actual results to optimize the marketing strategy. Adjustments can also be made based on specific business scenarios and data. In practice, other factors may also need to be considered, such as seasonal effects and competitor activities.

[0147] As an optional approach, the first sub-variation parameter is input into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel. This includes: when the channel marketing effects between each channel do not affect each other, using a saturation transformation function to transform the first sub-variation parameter corresponding to each channel into the second sub-transformation parameter corresponding to each channel; when the channel marketing effects between each channel affect each other, using a target hyperparameter determined by the channel characteristics to transform the first sub-variation parameter corresponding to each channel into the second sub-transformation parameter corresponding to each channel.

[0148] It should be noted that the embodiments in this application include two scenarios: one where the marketing effects of each channel do not affect each other, and the other where the marketing effects of each channel affect each other. In these two scenarios, different methods can be used to transform the first sub-transformation parameter to the second sub-transformation parameter.

[0149] For example, if the marketing effectiveness of different channels does not affect each other, the marketing effectiveness of each channel can be considered independently. A sigmoid function is a mathematical function that maps a real number or a set of real numbers to the interval (0,1). A typical application of this function is in logistic regression, where it is used to convert the output of a linear function into probability values. In this case, the sigmoid function can be used to linearly transform the first sub-parameter into the second sub-parameter without affecting the marketing effectiveness of other channels.

[0150] For example, in situations where the marketing effectiveness of different channels influences each other, the marketing performance of each channel will be affected by the marketing performance of other channels. To address this issue, a target hyperparameter, determined by the characteristics of each channel, can be used. The target hyperparameter is an additional parameter that can be used to adjust marketing strategies to accommodate the interactions between different channels. In practice, it may be necessary to determine the optimal value of the target hyperparameter through experimentation or data analysis to achieve the best marketing results across all channels.

[0151] As an optional approach, the evaluation results are determined by comparing predicted marketing results with historical results data. This includes: determining the predicted marketing results for each channel based on predicted marketing results, channel marketing investment, and channel elasticity parameters; determining the actual channel marketing results for each channel based on historical investment data and historical results data; and determining the evaluation results for each channel based on the magnitude of the predicted marketing results and actual channel marketing results.

[0152] For example, to evaluate the marketing effectiveness of each channel, the following steps can be taken:

[0153] S1 collects historical investment data (such as advertising expenses, promotional activity costs, etc.) and historical performance data (such as sales volume, number of customers, etc.) for each channel.

[0154] S2 calculates the predicted marketing effect for each channel based on the predicted marketing effect, channel marketing investment, and channel elasticity parameters. This can be achieved by constructing a predictive model, such as regression analysis or time series analysis.

[0155] S3 uses historical input and performance data to calculate the true channel marketing effectiveness for each channel. This can be achieved by calculating the return on investment (ROI) or other relevant metrics.

[0156] S4 compares the predicted marketing results for each channel with the actual marketing results for each channel.

[0157] S5 determines the evaluation result for each channel based on the relationship between predicted marketing results and actual channel marketing results. This can be achieved using methods including, but not limited to:

[0158] If the predicted marketing results are higher than the actual channel marketing results, it indicates that the channel has the potential to improve marketing effectiveness, and investment can be increased or marketing strategies optimized. If the predicted marketing results are similar to the actual channel marketing results, it indicates that the channel's marketing results are in line with expectations, and existing investment and strategies can be maintained. If the predicted marketing results are lower than the actual channel marketing results, it indicates that the channel's marketing results may be overestimated, and marketing strategies or investment need to be reassessed. Based on the assessment results, corresponding optimization strategies should be developed for each channel to improve overall marketing efficiency.

[0159] As an optional approach, the actual channel marketing effect for each channel is determined based on historical input and effect data. This includes: determining the first historical input and effect data for the target channel that did not participate in the campaign, and the second historical input and effect data for the target channel that did participate, where the target channel refers to any one of the channels; training an initial marketing effect prediction model based on the first historical input and effect data to obtain a target marketing effect prediction model; inputting the second historical input data into the counterfactual marketing effect prediction to determine the counterfactual predicted marketing effect for the target channel; and determining the actual channel marketing effect for the target channel based on the second historical effect data and the counterfactual predicted marketing effect.

[0160] Optionally, in this embodiment, the target channel refers to any one of the channels. The steps for determining the target channel's actual marketing effectiveness may include, but are not limited to:

[0161] S1. Collect historical data: First, it is necessary to collect historical investment data and historical performance data of the target channels that did not participate in the campaign, as well as historical investment data and historical performance data of the target channels that did participate in the campaign.

[0162] S2. Training the prediction model: Using historical investment data and historical performance data that were not involved in the campaign, train an initial marketing performance prediction model to obtain the target marketing performance prediction model.

[0163] S3. Predicting Marketing Effectiveness: Input the second historical investment data of the campaign into the target marketing effectiveness prediction model to predict the counterfactual predictive marketing effectiveness of the target channel. This counterfactual predictive marketing effectiveness can be determined, including but not limited to, through the following methods:

[0164] First, the second historical input data is applied to the target marketing effect prediction model to obtain the counterfactual predicted marketing effect y_pred. By comparing the second historical effect data y_observed with the counterfactual result value y_pred predicted by the model, the true channel marketing effect lift value of the channel is calculated.

[0165] S4. Determine True Marketing Efficiency: Finally, based on the second historical performance data of the campaign and the predicted channel marketing performance, calculate the true marketing efficiency of the target channel to evaluate the marketing performance of the target channel and determine its true marketing efficiency. Each channel may have its own true channel marketing efficiency, including but not limited to its own. The formula for calculating the true channel marketing efficiency lift value is as follows:

[0166] Lift = y_observed - y_pred.

[0167] As an optional approach, the predicted marketing effect for each channel is determined based on the predicted marketing effect, channel marketing investment, and channel elasticity parameters. This includes: determining the target channel marketing investment and target channel elasticity parameters for the target channel based on the channel marketing investment and channel elasticity parameters; determining a first intermediate value using the target channel marketing investment as the base and the target channel elasticity parameters as the exponent; determining a second intermediate value based on the ratio of the predicted marketing effect to the first intermediate value; determining the target predicted marketing effect for the target channel based on the difference between the predicted marketing effect and the second intermediate value; and determining the evaluation result for each channel based on the relationship between the predicted marketing effect and the actual channel marketing effect. This includes: if the actual channel marketing effect is less than or equal to the target predicted marketing effect, determining the target evaluation result for the target channel indicates that the marketing efficiency of the target channel meets the preset conditions; if the actual channel marketing effect is greater than the target predicted effect, determining the target evaluation result indicates that the marketing efficiency of the target channel does not meet the preset conditions.

[0168] For example, the marketing investment and elasticity parameters of the target channel need to be determined first, and then a first median value is calculated. Next, the predicted marketing effect of the target channel is compared with the first median value to obtain a second median value. The difference between the predicted marketing effect of the target channel and the second median value is then determined as the target predicted marketing effect of the target channel. Finally, the evaluation result for each channel is determined based on the relationship between the predicted marketing effect and the actual channel marketing effect, in order to determine whether the marketing efficiency of the target channel meets the preset conditions.

[0169] In an exemplary embodiment, firstly, it is necessary to determine the target channel marketing investment and target channel elasticity parameter corresponding to the target channel. The target channel marketing investment is used as the base and the target channel elasticity parameter is used as the exponent to determine the first intermediate value. Next, the ratio of the target predicted marketing effect corresponding to the target channel to the first intermediate value is determined as the second intermediate value to establish a relative effect value.

[0170] Next, the difference between the target predicted marketing effect and the second median value is determined as the target predicted marketing effect for the target channel. This step is to determine the predicted efficiency value. For example, if the actual channel marketing effect is less than or equal to the target predicted marketing effect, then the marketing efficiency of the target channel meets the preset conditions; or, if the actual channel marketing effect is greater than the target predicted marketing effect, then the marketing efficiency of the target channel does not meet the preset conditions. Finally, the evaluation result for each channel is determined based on the relationship between the predicted marketing effect and the actual channel marketing effect. A reference value can be established by analyzing the marketing investment and elasticity parameters of the target channel, and the marketing efficiency of each channel can be determined by comparing the predicted marketing effect and the actual channel marketing effect, thus more accurately evaluating the marketing effect of each channel and better formulating marketing strategies and allocating resources.

[0171] In summary, this method and process involves using data analysis and comparison to determine the marketing efficiency of target channels. This approach helps businesses better understand the performance of each channel and make corresponding adjustments and optimizations.

[0172] As an optional approach, feature extraction is performed on historical input data to determine the explanatory and control variables corresponding to the initial channel combination model. This includes: determining the correlation coefficient between channel marketing input and historical performance data for each channel in a set of channels based on historical input data, where the correlation coefficient is used to represent the correlation between each channel and historical performance data; and performing feature extraction on channels whose correlation coefficients meet preset sorting conditions to determine the explanatory and control variables corresponding to the initial channel combination model.

[0173] It's important to clarify that, firstly, the correlation coefficient is a statistical method used to measure the degree of correlation between two variables. The correlation coefficient is used to measure the correlation between the channel transformation value of channel marketing investment for each channel and historical performance data. If the correlation coefficient between two variables is close to 1, it indicates a strong positive correlation; if the correlation coefficient is close to -1, it indicates a strong negative correlation; and if the correlation coefficient is close to 0, it indicates that there is essentially no correlation between them.

[0174] Secondly, based on the preset ranking criteria of the correlation coefficients, the channels can be ranked to identify those with the highest correlation to historical performance data. This allows us to determine which channels have a greater impact on historical performance data, enabling us to focus our attention and analysis on these channels. Once the channels with the highest correlation to historical performance data are identified, feature extraction can be performed. Feature extraction refers to extracting useful features from a large amount of data to build a model. Specifically, feature extraction extracts the factors that influence historical performance data from the most correlated channels, serving as explanatory and control variables. Through feature extraction, the explanatory and control variables corresponding to the initial channel combination model can be determined. Explanatory variables explain changes in the dependent variable in the model, while control variables control other factors that affect the dependent variable. By determining these variables, a preliminary channel combination model can be constructed to analyze and predict the impact of each channel on historical performance data. By using correlation coefficients and feature extraction to determine the explanatory and control variables in the channel combination model, we can analyze and predict the marketing investment and historical performance data of each channel.

[0175] As an optional approach, after determining the initial channel combination model as the target channel combination model, assuming the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the method further includes: obtaining first expected input data, wherein the first expected input data includes the total expected input to a group of channels; inputting the first expected input data into the target channel combination model, and allocating the first expected input data based on the marketing efficiency of each channel in the group of channels through the target channel combination model to generate first expected effect data, wherein the first expected effect data represents the maximum marketing effect that can be obtained by marketing according to the first expected input data.

[0176] In an exemplary embodiment, it is assumed that the above-mentioned channel group consists of three channels, namely channel A, channel B, and channel C. The marketing efficiency A of channel A, the marketing efficiency B of channel B, and the marketing efficiency C of channel C are obtained through the above evaluation results. Furthermore, the marketing efficiency A, marketing efficiency B, and marketing efficiency C all meet the above-mentioned preset conditions. At this time, the training of the initial channel combination model is completed, and the initial channel combination model is determined as the target channel combination model.

[0177] Furthermore, the aforementioned first expected investment data can be allocated to Channel A, Channel B, and Channel C. The first expected investment data can be understood as the total resources or budget expected to be invested, which may include, but is not limited to, money, time, and manpower. For example, the first expected investment data may include, but is not limited to, social media advertising budget: the budget for placing ads on social media platforms; email marketing budget: the budget for sending marketing emails and maintaining email lists; search engine marketing budget: the budget for purchasing keyword ads and optimizing search engine rankings; television advertising budget: the budget for placing ads on television; and outdoor advertising budget: the budget for placing ads on outdoor media (such as bus stops, billboards, etc.). The first expected input data is then input into the target channel combination model. This model yields the marketing efficiency of each of the three channels: Channel A, Channel B, and Channel C. Based on the respective marketing efficiencies of Channel A, Channel B, and Channel C, the first expected input data is allocated accordingly. For example, if Channel A has the highest marketing efficiency and Channel C has the lowest, the first expected input data can be distributed among the channels according to their marketing efficiency. For instance, 50% of the first expected input data can be allocated to Channel A, 35% to Channel B, and 15% to Channel C. The first expected effect data is then generated based on the marketing performance of Channels A, B, and C. This first expected effect data can be understood as the optimal marketing effect achievable with the best possible strategy for the first expected input data. For example, the first expected effect data may include, but is not limited to: expected sales revenue (the expected sales revenue achievable through a set of channels); expected customer numbers (the expected number of customers attracted through a set of channels); expected brand awareness (the expected increase in brand awareness through a set of channels); and expected market share (the expected market share obtained through a set of channels).

[0178] As an optional approach, after determining the initial channel combination model as the target channel combination model, assuming the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the method further includes: obtaining second expected effect data, wherein the second expected effect data includes the total expected effect to be obtained from a group of channels; inputting the second expected effect data into the target channel combination model, and determining the second expected input data allocated to the group of channels based on the marketing efficiency of each channel in the group of channels through the target channel combination model, wherein the second expected input data represents the minimum marketing input required to obtain the second expected effect data.

[0179] In an exemplary embodiment, assuming the aforementioned channel set consists of three channels, namely Channel A, Channel B, and Channel C, the aforementioned second expected effect data can be understood as the expected marketing effect that can be obtained by distributing data to Channel A, Channel B, and Channel C. For example, the second expected effect data may include, but is not limited to, expected sales revenue: the expected sales revenue that can be achieved through the channel set; expected number of customers: the expected number of customers that can be attracted through the channel set; expected brand awareness: the expected brand awareness that can be increased through the channel set; and expected market share: the expected market share that can be obtained through the channel set.

[0180] Furthermore, the second expected effect data can be used as input to the target channel mix model. For example, the second expected effect data represents the expectation of achieving 80% market share through the aforementioned group of channels. After processing by the target channel mix model, the model output is the data required for each channel in the aforementioned group. For example, considering the marketing efficiency of the channels, the data required for channel A is 100, channel B is 200, and channel C is 300. Therefore, the second expected investment data is 300. The required investment data is determined by the data required for channel A, channel B, and channel C. In other words, the target channel mix model can use the second expected effect data as input and output the aforementioned second expected investment data. This second expected investment data can be understood as the expected investment... The minimum total resources or budget for investment can include, but is not limited to, money, time, and manpower. For example, the second expected investment data can include, but is not limited to, social media advertising budget: the budget for advertising on social media platforms; email marketing budget: the budget for sending marketing emails and maintaining email lists; search engine marketing budget: the budget for purchasing keyword advertising and optimizing search engine rankings; television advertising budget: the budget for advertising on television; and outdoor advertising budget: the budget for advertising on outdoor media (such as bus stops, billboards, etc.). The model outputs the minimum second expected investment data that needs to be allocated to different channels in the above set of channels. For example, 50% of the second expected investment data can be allocated to channel A, 40% to channel B, and 10% to channel C.

[0181] It is understood that in the specific embodiments of this application, data such as user information are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0182] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0183] According to another aspect of the embodiments of this application, a data processing apparatus for implementing the above-described data processing method is also provided. For example... Figure 16 As shown, the device includes:

[0184] The acquisition module 1602 is used to acquire historical investment data and historical performance data associated with the target application, wherein the target application is configured to conduct marketing activities through a set of channels, the historical investment data represents the marketing investment allocated to a set of channels, and the historical performance data represents the marketing performance generated by the set of channels conducting marketing activities according to the historical investment data.

[0185] Module 1604 is used to build an initial channel combination model based on historical investment data and historical effect data. The initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities carried out by each channel on the channel marketing effect in the future period. The second transformation sub-model is used to determine the degree to which the increase in channel marketing effect decreases as channel marketing investment increases.

[0186] Analysis module 1606 is used to conduct quasi-experimental analysis on the initial channel mix model to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel. The quasi-experimental analysis includes determining the predicted marketing effect of each channel based on historical investment data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and historical effect data.

[0187] The determination module 1608 is used to determine the initial channel combination model as the target channel combination model when the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions.

[0188] As an optional approach, the device is used to construct an initial channel combination model based on historical input data and historical effect data in the following manner: performing feature extraction on the historical input data to determine the explanatory variables and control variables corresponding to the initial channel combination model, wherein the explanatory variables represent the independent variables of the initial channel combination model, and the control variables represent the constants of the initial channel combination model; performing Bayesian estimation on the model parameters of the initial channel combination model based on the explanatory variables and control variables to determine the initial channel combination model.

[0189] As an optional approach, the device is used to estimate the model parameters of the initial channel combination model based on explanatory and control variables in the following manner to determine the initial channel combination model: obtaining the variable types of explanatory and control variables; pre-setting the initial model parameters according to the variable types, wherein, when the variable type includes media variables, the prior distribution corresponding to the initial model parameters is a non-negative distribution; when the variable type includes the decay rate of channel marketing effectiveness of different channels, the prior distribution corresponding to the initial model parameters is determined by the decay rate; when the variable type includes the peak value of channel marketing effectiveness of different channels, the prior distribution corresponding to the initial model parameters is determined by the peak value; determining the target likelihood function based on historical input data and historical effect data, wherein the target likelihood function is used to determine the probability distribution of historical effect data obtained based on historical input data when the initial model parameters are set; updating the target model parameters according to the target likelihood function and the initial model parameters to determine the initial channel combination model, wherein the target model parameters represent the posterior distribution determined according to the target likelihood function and the initial model parameters.

[0190] As an optional approach, the device is used to perform a quasi-experimental analysis on the initial channel mix model in the following manner to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel: determining a first set of media features and a second set of media features corresponding to each channel, wherein the first set of media features represents media features that belong to the explanatory and control variables, and the second set of media features represents media features that do not belong to the explanatory and control variables, with one media feature corresponding to one channel; inputting the marketing investment of the channels corresponding to the first set of media features from the historical investment data into the first transformation sub-model, and determining the first channel corresponding to each channel of the first set of media features. The first sub-transformation parameter is input into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel. The product of the second sub-transformation parameter corresponding to each channel and the channel elasticity parameter corresponding to each channel is determined as the channel transformation value of the channel marketing investment for each channel. Here, the channel elasticity parameter represents the magnitude of change in channel marketing effect as the channel marketing investment changes. The initial channel combination model is adjusted to obtain the target channel combination model. The sum of the channel transformation values ​​of the channel marketing investment corresponding to each channel is determined as the predicted marketing effect. The evaluation result is determined by comparing the predicted marketing effect with historical effect data.

[0191] As an optional solution, the device is used to input the channel marketing investment corresponding to the first set of media characteristics from the historical investment data into the first transformation sub-model in the following manner to determine the first sub-transformation parameter corresponding to each channel: obtaining the channel marketing investment corresponding to the first set of media characteristics for L unit statistical periods for each channel, where L represents the total duration of marketing influence of a channel; obtaining the attenuation parameter and the peak influence parameter pre-set for each channel, where the attenuation parameter represents the rate of attenuation of the channel marketing effect corresponding to a channel over time, and the peak influence parameter represents the lag time relative to the initial placement when the channel marketing effect corresponding to a channel reaches its peak; and determining the first sub-transformation parameter corresponding to each channel based on the channel marketing investment, attenuation parameter, and peak influence parameter corresponding to the first set of media characteristics.

[0192] As an optional approach, the device is used to input the first sub-variation parameter corresponding to each channel into the second transformation sub-model in the following manner to determine the second sub-transformation parameter corresponding to each channel: when the channel marketing effects between each channel do not affect each other, the first sub-variation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using a saturated transformation function; when the channel marketing effects between each channel affect each other, the first sub-variation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using a target hyperparameter determined by the channel characteristics.

[0193] As an optional solution, the device is used to determine the evaluation result by comparing predicted marketing effect and historical effect data in the following manner: determining the predicted marketing effect for each channel based on predicted marketing effect, channel marketing investment, and channel elasticity parameters; determining the actual channel marketing effect for each channel based on historical investment data and historical effect data; and determining the evaluation result for each channel based on the magnitude relationship between the predicted marketing effect and the actual channel marketing effect.

[0194] As an optional solution, the device is used to determine the actual channel marketing effect corresponding to each channel based on historical input data and historical effect data in the following manner: determining first historical input data and first historical effect data for which the target channel did not participate in the campaign, and second historical input data and second historical effect data for which the target channel participated in the campaign, wherein the target channel represents any one of the channels; training an initial marketing effect prediction model based on the first historical input data and first historical effect data to obtain a target marketing effect prediction model; inputting the second historical input data into the target marketing effect prediction model to determine the counterfactual predicted marketing effect corresponding to the target channel; and determining the actual channel marketing effect of the target channel based on the second historical effect data and the counterfactual predicted marketing effect.

[0195] As an optional solution, the device is used to determine the predicted marketing effect for each channel based on the predicted marketing effect, channel marketing investment, and channel elasticity parameters in the following manner: Determine the target channel marketing investment and target channel elasticity parameters for the target channel based on the channel marketing investment and channel elasticity parameters; determine a first intermediate value using the target channel marketing investment as the base and the target channel elasticity parameters as the exponent; determine a second intermediate value based on the ratio of the predicted marketing effect to the first intermediate value; determine the target predicted marketing effect for the target channel based on the difference between the predicted marketing effect and the actual channel marketing effect in the following manner: If the actual channel marketing effect is less than or equal to the target predicted marketing effect, determine that the target evaluation result for the target channel indicates that the marketing efficiency of the target channel meets the preset conditions; if the actual channel marketing effect is greater than the target predicted effect, determine that the target evaluation result indicates that the marketing efficiency of the target channel does not meet the preset conditions.

[0196] As an optional approach, the device is used to perform feature extraction on historical input data to determine the explanatory and control variables corresponding to the initial channel combination model by: determining the correlation coefficient between the channel marketing input and historical performance data of each channel in a set of channels based on the historical input data, wherein the correlation coefficient is used to represent the correlation between each channel and the historical performance data; performing feature extraction on channels whose correlation coefficients meet preset sorting conditions to determine the explanatory and control variables corresponding to the initial channel combination model.

[0197] As an optional solution, the apparatus is further configured to: after determining the initial channel combination model as the target channel combination model, obtain first expected input data, wherein the first expected input data includes the total amount of expected input to a group of channels; input the first expected input data into the target channel combination model, and allocate the first expected input data based on the marketing efficiency of each channel in the group of channels through the target channel combination model to generate first expected effect data, wherein the first expected effect data represents the maximum marketing effect that can be obtained by marketing according to the first expected input data.

[0198] As an optional embodiment, the apparatus is further configured to: after determining the initial channel combination model as the target channel combination model, obtain second expected effect data, wherein the second expected effect data includes the total expected effect to be obtained from a group of channels; input the second expected effect data into the target channel combination model, and determine the second expected input data allocated to the group of channels based on the marketing efficiency of each channel in the group of channels through the target channel combination model, wherein the second expected input data represents the minimum marketing input required to obtain the second expected effect data.

[0199] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0200] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0201] According to one aspect of this application, a computer program product is provided, the computer program product comprising a computer program.

[0202] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0203] Figure 17 A schematic block diagram of a computer system architecture for implementing an electronic device according to embodiments of the present application is shown.

[0204] It should be noted that, Figure 17 The computer system 1700 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0205] like Figure 17 As shown, the computer system 1700 includes a central processing unit (CPU) 1701, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1702 or programs loaded from storage section 1708 into random access memory (RAM) 1703. The RAM 1703 also stores various programs and data required for system operation. The CPU 1701, ROM 1702, and RAM 1703 are interconnected via a bus 1704. An input / output interface 1705 (I / O interface) is also connected to the bus 1704.

[0206] The following components are connected to the input / output interface 1705: an input section 1706 including a keyboard, mouse, etc.; an output section 1707 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1708 including a hard disk, etc.; and a communication section 1709 including a network interface card such as a local area network card, modem, etc. The communication section 1709 performs communication processing via a network such as the Internet. A drive 1710 is also connected to the input / output interface 1705 as needed. Removable media 1711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on the drive 1710 as needed so that computer programs read from them can be installed into the storage section 1708 as needed.

[0207] Specifically, according to embodiments of this application, the processes described in the various method flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1709, and / or installed from removable medium 1711. When the computer program is executed by central processing unit 1701, it performs various functions defined in the system of this application.

[0208] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 1709, and / or installed from removable media 1711. When the computer program is executed by central processing unit 1701, it performs various functions provided in the embodiments of this application.

[0209] According to another aspect of the embodiments of this application, an electronic device for implementing the above-described data processing method is also provided. This electronic device may be... Figure 1 The terminal device or server shown. This embodiment uses this electronic device as an example for illustration. Figure 18 As shown, the electronic device includes a memory 1802 and a processor 1804. The memory 1802 stores a computer program, and the processor 1804 is configured to execute the steps of any of the above method embodiments via the computer program.

[0210] Optionally, in this embodiment, the aforementioned electronic device may be located in at least one of a plurality of network devices in a computer network.

[0211] Optionally, in this embodiment, the processor may be configured to execute the methods in the embodiments of this application via a computer program.

[0212] Alternatively, as those skilled in the art will understand, Figure 18 The structure shown is for illustrative purposes only. Figure 18 This does not limit the structure of the aforementioned electronic devices. For example, the electronic device may also include components that are more... Figure 18 The more or fewer components shown (such as network interfaces, etc.), or having the same Figure 18 The different configurations shown.

[0213] The memory 1802 can be used to store software programs and modules, such as the program instructions / modules corresponding to the data processing methods and apparatus in this embodiment. The processor 1804 executes various functional applications and data processing by running the software programs and modules stored in the memory 1802, thereby implementing the aforementioned data processing methods. The memory 1802 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1802 may further include memory remotely located relative to the processor 1804, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Specifically, the memory 1802 may be used, but is not limited to, to store information such as historical input data, historical effect data, and predicted marketing effects. As an example, such as... Figure 18 As shown, the memory 1802 may include, but is not limited to, the acquisition module 1602, construction module 1604, analysis module 1606, and determination module 1608 in the data processing device. Furthermore, it may include, but is not limited to, other module units in the data processing device, which will not be described further in this example.

[0214] Optionally, the transmission device 1806 described above is used to receive or send data via a network. Specific examples of the network described above may include wired networks and wireless networks. In one example, the transmission device 1806 includes a Network Interface Controller (NIC), which can be connected to other network devices and a router via a network cable to communicate with the Internet or a local area network. In another example, the transmission device 1806 is a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0215] In addition, the aforementioned electronic device also includes: a display 1808 for displaying the aforementioned historical investment data, historical performance data, and predicted marketing performance; and a connection bus 1810 for connecting the various module components in the aforementioned electronic device.

[0216] In other embodiments, the aforementioned terminal device or server can be a node in a distributed system, wherein the distributed system can be a blockchain system, which is a distributed system formed by connecting multiple nodes through network communication. The nodes can form a peer-to-peer network, and any form of computing device, such as a server, terminal, or other electronic device, can become a node in the blockchain system by joining this peer-to-peer network.

[0217] According to one aspect of this application, a computer-readable storage medium is provided, wherein a processor of an electronic device reads computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the electronic device to perform a data processing method provided in various alternative implementations of the above-described data processing aspect.

[0218] Optionally, in this embodiment, the computer-readable storage medium described above may be configured to store methods for performing the embodiments of this application.

[0219] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0220] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0221] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.

[0222] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0223] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0224] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0225] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0226] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A data processing method, characterized in that, include: Acquire historical investment data and historical performance data associated with a target application, wherein the target application is configured to conduct marketing campaigns through a set of channels, the historical investment data represents the marketing investment allocated to the set of channels, and the historical performance data represents the marketing performance generated by the set of channels conducting marketing campaigns according to the historical investment data; An initial channel combination model is constructed based on the historical investment data and the historical effect data. The initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities carried out by each channel on the channel marketing effect in the future period. The second transformation sub-model is used to determine the degree to which the increase in channel marketing effect decreases as channel marketing investment increases. A quasi-experimental analysis is conducted on the initial channel combination model to determine the evaluation results of the marketing efficiency of the initial channel combination model for each channel. The quasi-experimental analysis includes determining the predicted marketing effect based on the historical investment data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and the historical effect data. If the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the initial channel combination model is determined as the target channel combination model.

2. The method according to claim 1, characterized in that, The step of constructing an initial channel combination model based on the historical input data and the historical performance data includes: Perform feature extraction on the historical input data to determine the explanatory and control variables corresponding to the initial channel combination model, wherein the explanatory variables represent the independent variables of the initial channel combination model, and the control variables represent the constants of the initial channel combination model; Based on the explanatory variables and the control variables, Bayesian estimation is performed on the model parameters of the initial channel combination model to determine the initial channel combination model.

3. The method according to claim 2, characterized in that, The step of estimating the model parameters of the initial channel combination model based on the explanatory variables and the control variables, and determining the initial channel combination model, includes: Obtain the variable types of the explanatory variables and the control variables; Initial model parameters are pre-set according to the variable type. When the variable type includes media variables, the prior distribution of the initial model parameters is a non-negative distribution. When the variable type includes the decay rate of channel marketing effect of different channels, the prior distribution of the initial model parameters is determined by the decay rate. When the variable type includes the peak value of channel marketing effect of different channels, the prior distribution of the initial model parameters is determined by the peak value. A target likelihood function is determined based on the historical input data and the historical effect data, wherein the target likelihood function is used to determine the probability distribution of obtaining the historical effect data based on the historical input data, given the initial model parameters; The target model parameters are updated based on the target likelihood function and the initial model parameters to determine the initial channel combination model, wherein the target model parameters represent the posterior distribution determined based on the target likelihood function and the initial model parameters.

4. The method according to claim 2, characterized in that, The quasi-experimental analysis of the initial channel mix model to determine the evaluation results of the marketing efficiency of the initial channel mix model for each channel includes: Determine a first set of media features and a second set of media features corresponding to each channel, wherein the first set of media features represents media features that belong to the explanatory variables and the control variables, and the second set of media features represents media features that do not belong to the explanatory variables and the control variables, and one media feature corresponds to one channel; Input the channel marketing investment corresponding to the first group of media features from the historical investment data into the first transformation sub-model to determine the first sub-transformation parameter corresponding to each channel corresponding to the first group of media features. Input the first sub-transformation parameter corresponding to each channel into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel; The product of the second sub-transformation parameter corresponding to each channel and the channel elasticity parameter corresponding to each channel is determined as the channel transformation value of the channel marketing investment corresponding to each channel. The channel elasticity parameter represents the magnitude of change in channel marketing effect as the channel marketing investment changes. The initial channel combination model is obtained by adjusting the channel elasticity parameter to obtain the target channel combination model. The sum of the channel transformation values ​​corresponding to each channel is determined as the predicted marketing effect; The evaluation result is determined by comparing the predicted marketing effect with the historical effect data.

5. The method according to claim 4, characterized in that, The step of inputting the channel marketing investment corresponding to the first group of media characteristics from the historical investment data into the first transformation sub-model to determine the first sub-transformation parameters corresponding to each channel includes: Obtain the channel marketing investment corresponding to the first set of media characteristics for L statistical periods for each channel, where L represents the total duration of marketing influence for a channel; Obtain the attenuation parameter and peak impact parameter pre-set for each channel, wherein the attenuation parameter represents the rate at which the channel marketing effect of a channel decreases over time, and the peak impact parameter represents the lag time relative to the initial launch when the channel marketing effect of a channel reaches its peak. The first sub-transformation parameter corresponding to each channel is determined based on the channel marketing investment, the attenuation parameter, and the peak impact parameter corresponding to the first group of media characteristics.

6. The method according to claim 4, characterized in that, The step of inputting the first sub-transformation parameter corresponding to each channel into the second transformation sub-model to determine the second sub-transformation parameter corresponding to each channel includes: When the channel marketing effects of each channel do not affect each other, the first sub-transformation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using a saturation transformation function; When the channel marketing effects of each channel affect each other, the first sub-transformation parameter corresponding to each channel is transformed into the second sub-transformation parameter corresponding to each channel using the target hyperparameter determined by the channel characteristics.

7. The method according to claim 4, characterized in that, The step of comparing the predicted marketing effect with the historical effect data to determine the evaluation result includes: The predicted marketing effect for each channel is determined based on the predicted marketing effect, the channel marketing investment, and the channel elasticity parameter. The actual channel marketing effect corresponding to each channel is determined based on the historical input data and the historical effect data. The evaluation result for each channel is determined based on the relationship between the predicted marketing effect and the actual channel marketing effect.

8. The method according to claim 7, characterized in that, The step of determining the actual channel marketing effect for each channel based on the historical investment data and the historical effect data includes: Based on the historical investment data and the historical effect data, determine the first historical investment data and the first historical effect data of the target channel not participating in the campaign, and the second historical investment data and the second historical effect data of the target channel participating in the campaign, wherein the target channel refers to any one of the channels; The initial marketing effect prediction model is trained based on the first historical input data and the first historical effect data to obtain the target marketing effect prediction model. Input the second historical input data into the target marketing effect prediction model to determine the counterfactual predicted marketing effect corresponding to the target channel; The true channel marketing effect of the target channel is determined based on the second historical performance data and the counterfactual predicted marketing effect.

9. The method according to claim 8, characterized in that, The step of determining the predicted marketing effect for each channel based on the predicted marketing effect, the channel marketing investment, and the channel elasticity parameter includes: determining the target channel marketing investment and the target channel elasticity parameter for the target channel based on the channel marketing investment and the channel elasticity parameter; determining a first intermediate value with the target channel marketing investment as the base and the target channel elasticity parameter as the exponent; determining a second intermediate value as the ratio of the predicted marketing effect to the first intermediate value; and determining the difference between the predicted marketing effect and the second intermediate value as the target predicted marketing effect for the target channel. The step of determining the evaluation result for each channel based on the relationship between the predicted marketing effect and the actual channel marketing effect includes: when the actual channel marketing effect is less than or equal to the target predicted marketing effect, determining that the target evaluation result for the target channel indicates that the marketing efficiency of the target channel meets a preset condition; and when the actual channel marketing effect is greater than the target predicted marketing effect, determining that the target evaluation result indicates that the marketing efficiency of the target channel does not meet a preset condition.

10. The method according to claim 2, characterized in that, The step of performing feature extraction on the historical input data to determine the explanatory and control variables corresponding to the initial channel combination model includes: Based on the historical investment data, the correlation coefficient between the channel marketing investment corresponding to each channel in the group of channels and the historical performance data is determined, wherein the correlation coefficient is used to represent the correlation between each channel and the historical performance data; For channels whose correlation coefficients meet preset sorting conditions, feature extraction is performed to determine the explanatory and control variables corresponding to the initial channel combination model.

11. The method according to claim 1, characterized in that, After determining the initial channel combination model as the target channel combination model when the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the method further includes: Obtain first expected investment data, wherein the first expected investment data includes the total amount expected to be invested in the group of channels; The first expected input data is input into the target channel combination model. The target channel combination model allocates the first expected input data based on the marketing efficiency of each channel in the group of channels to generate the first expected effect data. The first expected effect data represents the maximum marketing effect that can be obtained by marketing according to the first expected input data.

12. The method according to claim 1, characterized in that, After determining the initial channel combination model as the target channel combination model when the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions, the method further includes: Obtain second expected effect data, wherein the second expected effect data includes the total expected effect obtained from the set of channels; The second expected effect data is input into the target channel combination model. The target channel combination model determines the second expected investment data allocated to the group of channels based on the marketing efficiency of each channel in the group of channels. The second expected investment data represents the minimum marketing investment required to obtain the second expected effect data.

13. A data processing apparatus, characterized in that, include: The acquisition module is used to acquire historical investment data and historical performance data associated with a target application, wherein the target application is configured to conduct marketing activities through a set of channels, the historical investment data represents the marketing investment allocated to the set of channels, and the historical performance data represents the marketing performance generated by the set of channels conducting marketing activities according to the historical investment data. A construction module is used to construct an initial channel combination model based on the historical investment data and the historical effect data. The initial channel combination model includes a first transformation sub-model and a second transformation sub-model. The first transformation sub-model is used to determine the cumulative impact of marketing activities carried out by each channel on the channel marketing effect in the future period. The second transformation sub-model is used to determine the degree to which the increase in channel marketing effect decreases as channel marketing investment increases. An analysis module is used to perform quasi-experimental analysis on the initial channel combination model to determine the evaluation results of the marketing efficiency of the initial channel combination model for each channel. The quasi-experimental analysis includes determining the predicted marketing effect of each channel based on the historical investment data, and evaluating the marketing efficiency of each channel based on the predicted marketing effect and the historical effect data. The determination module is used to determine the initial channel combination model as the target channel combination model when the evaluation results indicate that the marketing efficiency of each channel meets the preset conditions.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein the computer program can be executed by an electronic device to perform the method described in any one of claims 1 to 12.

15. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1 to 12.

16. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 12 through the computer program.

Images