Advertisement overflow value prediction method, device and equipment

Abstract

The invention discloses a method, a device and equipment for predicting an advertisement overflow value. The method comprises the following steps: acquiring a front-end advertisement data set; preprocessing the front-end advertisement data set to obtain a transition advertisement data set; performing inspection processing on the transition advertisement data set to obtain a target advertisement data set; inputting the target advertisement data set into a first prediction layer of an advertisement overflow value prediction model for processing to obtain a first layer processing result; and inputting the first layer processing result and a part of the target advertisement data set into a second prediction layer of an advertisement overflow value prediction model to obtain a target advertisement overflow value. According to the scheme provided by the invention, the specific overflow value of the advertisement can be directly predicted, so that an advertisement putting person can more finely adjust budget allocation and optimize a putting strategy according to the specific overflow value, and effective utilization of resources is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of advertisement prediction, and in particular to a method, device and equipment for predicting advertisement spillover value. Background Art

[0002] In the current digital advertising ecosystem, artificial intelligence has been widely used in advertising effect prediction. The current advertising prediction methods are divided into prediction of single platform indicators and prediction of cross-platform data. Among them, the prediction of single platform indicators cannot fully capture the real impact of advertising materials, resulting in the accuracy and reliability of the prediction results being limited; and the existing cross-platform data prediction can only predict whether there is spillover, but cannot predict the specific spillover value, and cannot meet the needs of precise advertising delivery. Summary of the invention

[0003] The purpose of the present invention is to provide a method, device and equipment for predicting advertising spillover value in order to solve the above-mentioned problem, as described in detail below.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] The present invention provides a method for predicting advertising spillover value, comprising:

[0006] Get the front-end advertising dataset;

[0007] Preprocessing the front-end advertisement data set to obtain a transition advertisement data set;

[0008] Performing inspection processing on the transition advertisement data set to obtain a target advertisement data set;

[0009] Inputting the target advertisement data set into the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first layer processing result;

[0010] The first layer processing results and part of the target advertising data set are input into the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value; wherein the first prediction layer of the advertising spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertising spillover value prediction model is trained according to a second preset network model.

[0011] Optionally, preprocessing the front-end advertisement data set to obtain a transition advertisement data set includes:

[0012] Performing gap filling processing on the front-end advertisement data set to obtain a first pre-processed data set;

[0013] Performing outlier processing on the first preprocessed data set to obtain a second preprocessed data set;

[0014] Normalize the second preprocessed data set to obtain a transitional advertisement data set.

[0015] Optionally, perform a test process on the transitional advertisement data set to obtain a target advertisement data set, including:

[0016] Perform a unit root test process on the transitional advertisement data set to obtain a first test result;

[0017] Perform an autocorrelation test process on the first test result to obtain a target advertisement data set.

[0018] Optionally, input the target advertisement data set into the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first-layer processing result, including:

[0019] Input the target advertisement data set into the first processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first processing result;

[0020] Input the first processing result into the second processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a second processing result;

[0021] Input the second processing result into the third processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first-layer processing result.

[0022] Optionally, it further includes:

[0023] Process the target advertisement data set to obtain an autocorrelation graph;

[0024] Determine the potential order of the first prediction layer of the advertisement spillover value prediction model according to the autocorrelation graph.

[0025] Optionally, process the target advertisement data set to obtain an autocorrelation graph, including:

[0026] Perform an autocorrelation process on the target advertisement data set to obtain a first correlation graph;

[0027] Perform a partial autocorrelation process on the target advertisement data set to obtain a second correlation graph.

[0028] Optionally, input the first-layer processing result and a part of the target advertisement data set into the second prediction layer of the advertisement spillover value prediction model to obtain a target advertisement spillover value, including:

[0029] Input the first-layer processing result and a part of the target advertisement data set into the input layer of the second prediction layer of the advertisement spillover value prediction model to obtain an input layer result;

[0030] Input the input layer structure into the hidden layer of the second prediction layer of the advertising spillover value prediction model to obtain the hidden layer result;

[0031] Input the hidden layer result into the output layer of the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value.

[0032] The present invention also provides a prediction device for advertising spillover value, including:

[0033] An acquisition module, configured to acquire a front-end advertising data set;

[0034] A processing module, configured to preprocess the front-end advertising data set to obtain a transition advertising data set; perform inspection processing on the transition advertising data set to obtain a target advertising data set; input the target advertising data set into the first prediction layer of the advertising spillover value prediction model for processing to obtain a first-layer processing result; input the first-layer processing result and a part of the target advertising data set into the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value; wherein, the first prediction layer of the advertising spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertising spillover value prediction model is trained according to a second preset network model.

[0035] The present invention also provides a computing device, including: a processor and a memory storing a computer program, and when the computer program is run by the processor, it executes the method as described above.

[0036] The present invention also provides a computer-readable storage medium storing instructions, and when the instructions are run on a computer, the computer is caused to execute the method as described above.

[0037] The above solution of the present invention has at least the following beneficial effects:

[0038] The above solution of the present invention, by acquiring a front-end advertising data set; preprocessing the front-end advertising data set to obtain a transition advertising data set; performing inspection processing on the transition advertising data set to obtain a target advertising data set; inputting the target advertising data set into the first prediction layer of the advertising spillover value prediction model for processing to obtain a first-layer processing result; inputting the first-layer processing result and a part of the target advertising data set into the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value; can directly predict the specific spillover value of the advertisement, which is convenient for advertisers to more finely adjust the budget allocation and optimize the placement strategy according to the specific spillover value, and ensure the effective utilization of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flowchart of a prediction method for advertising spillover value provided by an embodiment of the present invention;

[0040] Figure 2 is a flowchart of the prediction process of the advertisement spillover value provided by an embodiment of the present invention;

[0041] Figure 3 is a module diagram of the prediction device of the advertisement spillover value provided by an embodiment of the present invention. Detailed implementation manners

[0042] Hereinafter, exemplary embodiments of the present invention will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be completely conveyed to those skilled in the art.

[0043] As Figure 1 shown, an embodiment of the present invention provides a method for predicting an advertisement spillover value, including:

[0044] Step 11, obtaining a front-end advertisement data set;

[0045] Step 12, preprocessing the front-end advertisement data set to obtain a transitional advertisement data set;

[0046] Step 13, performing inspection processing on the transitional advertisement data set to obtain a target advertisement data set;

[0047] Step 14, inputting the target advertisement data set into a first prediction layer of an advertisement spillover value prediction model for processing to obtain a first-layer processing result;

[0048] Step 15, inputting the first-layer processing result and a part of the target advertisement data set into a second prediction layer of the advertisement spillover value prediction model to obtain a target advertisement spillover value; wherein, the first prediction layer of the advertisement spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertisement spillover value prediction model is trained according to a second preset network model.

[0049] In this embodiment, by obtaining a front-end advertisement data set; preprocessing the front-end advertisement data set to obtain a transitional advertisement data set; performing inspection processing on the transitional advertisement data set to obtain a target advertisement data set; inputting the target advertisement data set into a first prediction layer of an advertisement spillover value prediction model for processing to obtain a first-layer processing result; inputting the first-layer processing result and a part of the target advertisement data set into a second prediction layer of the advertisement spillover value prediction model to obtain a target advertisement spillover value; it is possible to directly predict the specific spillover value of the advertisement, which is convenient for advertisers to more finely adjust the budget allocation and optimize the placement strategy according to the specific spillover value, and ensure the effective utilization of resources.

[0050] In step 11, the front-end advertisement dataset includes, but is not limited to, features such as consumption, interaction, publisher, user, etc. of the advertisement, and the total spillover value generated by the advertisement every day.

[0051] In an alternative embodiment of the present invention, step 12 includes:

[0052] Step 121, perform a filling-in processing on the front-end advertisement dataset to obtain a first preprocessed dataset;

[0053] Step 122, perform an outlier processing on the first preprocessed dataset to obtain a second preprocessed dataset;

[0054] Step 123, perform a normalization processing on the second preprocessed dataset to obtain a transitional advertisement dataset.

[0055] In this embodiment, when performing a filling-in processing on the front-end advertisement dataset to obtain a first preprocessed dataset, specifically, for binary classification data, the mode filling method is adopted, that is, the category value with the highest frequency of occurrence is selected to supplement the missing value; for numerical data, especially continuous variables, the mean value is used for filling to maintain the overall statistical characteristics of the dataset;

[0056] When performing an outlier processing on the first preprocessed dataset to obtain a second preprocessed dataset, specifically, the interquartile range (IQR) method is adopted, and the data points below the first quartile (Q1) minus 1.5 times the IQR or above the third quartile (Q3) plus 1.5 times the IQR are defined as outliers. After the outliers are identified and confirmed, they are removed from the dataset to prevent them from having an adverse impact on subsequent statistical analysis;

[0057] When performing a normalization processing on the second preprocessed dataset to obtain a transitional advertisement dataset, specifically, the normalization processing can be normalization processing, etc., aiming to eliminate the differences in units and numerical scales between features, prevent over-reliance on certain features during the model training process, and ensure that all features are considered equally;

[0058] In this embodiment, feature engineering is simultaneously performed to enhance the representational ability of the data by calculating derivative features such as various costs and ratios.

[0059] In an alternative embodiment of the present invention, step 13 includes:

[0060] Step 131, perform a unit root test processing on the transitional advertisement dataset to obtain a first test result;

[0061] Step 132: Perform autocorrelation detection processing on the first detection result to obtain a target advertisement dataset.

[0062] In this embodiment, unit root detection processing is performed on the transitional advertisement dataset to obtain a first detection result. Specifically, based on an autoregressive model, it is tested whether there is a unit root in the time series y t to confirm whether the time series is a stationary process, ensure that the data used satisfies the stationarity assumption of the model, and improve the reliability of the analysis result and the accuracy of prediction; specifically:

[0063]

[0064] where y t represents the value of the time series; Δ represents the difference operator; α represents the constant term; β represents the time trend term; φ represents the key parameter of the unit root test; γ i represents the coefficient of the lag term; ∈ t represents the error term;

[0065] By performing the above test, when φ = 0, it indicates that there is a unit root in the time series, which means the sequence is non-stationary; when φ < 0, it indicates that there is no unit root in the time series and the sequence is stationary; if the sequence is non-stationary, it is necessary to difference the sequence to make it stationary;

[0066] Perform autocorrelation detection processing on the first detection result to obtain a target advertisement dataset. Specifically, perform multi-lag autocorrelation test on the stationary sequence:

[0067]

[0068] where y t represents the value of the time series; represents the mean of the time series; k represents the lag period; T represents the total number of samples; Q represents the test statistic; T represents the total number of samples; m represents the number of lag periods; r k represents the autocorrelation coefficient of the k-th lag period;

[0069] By comparing Q with the critical value of the chi-square distribution, it is judged whether the autocorrelation of the data is significant, and it is verified whether there is a meaningful time-dependent structure in the data, rather than just being manifested as white noise; if the data is indeed white noise, it indicates that the data lacks valuable information that can be extracted, and further analysis is not appropriate, and it is necessary to collect the advertisement dataset again.

[0070] In an alternative embodiment of the present invention, step 14 includes:

[0071] Step 141: Input the target advertisement data set into the first processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first processing result;

[0072] Step 142: Input the first processing result into the second processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a second processing result;

[0073] Step 143: Input the second processing result into the third processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first-layer processing result.

[0074] In this embodiment, the first prediction layer of the advertisement spillover value prediction model is trained according to a first preset network model, where the first preset network model can be an ARIMA model (Autoregressive Integrated Moving Average model);

[0075] Input the target advertisement data set into the first processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first processing result. Specifically, input the target advertisement data set into the autoregressive part of the trained ARIMA model for processing to obtain a first processing result;

[0076] Input the first processing result into the second processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a second processing result. Specifically, input the first processing result into the differencing part of the trained ARIMA model for processing to obtain a second processing result;

[0077] Input the second processing result into the third processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first-layer processing result. Specifically, input the second processing result into the moving average part of the trained ARIMA model to obtain a first-layer processing result;

[0078] The training process of the ARIMA model: Use the Maximum Likelihood Estimation (MLE) method to estimate the model parameters

[0079]

[0080] To ensure the fitting quality of the model, check whether the residuals conform to the white noise characteristics, that is, the residuals should be random variables with a mean of zero, a constant variance, and no autocorrelation; if the initially established model fails to adequately fit the data, the model order (p, d, q) needs to be adjusted according to the results of the residual analysis, and parameter estimation should be performed again until a satisfactory fitting effect is obtained to determine the best-fitting ARIMA model. To evaluate the prediction performance of the model, calculate various evaluation metrics on an independent test set, specifically, the Mean Absolute Error (MAE) and the Mean Squared Error (MSE):

[0081]

[0082] where n represents the number of samples; y i represents the actual value; represents the predicted value;

[0083] Based on the above evaluation results, further fine-tune the model structure or parameter settings, and repeat the above process until the expected prediction accuracy is met.

[0084] In an alternative embodiment of the present invention, step 14 further includes:

[0085] Step 1400, process the target advertising data set to obtain an autocorrelation graph;

[0086] Step 1401, determine the potential order of the first prediction layer of the advertising spillover value prediction model according to the autocorrelation graph;

[0087] Further, step 1400 includes:

[0088] Perform autocorrelation processing on the target advertising data set to obtain a first correlation graph;

[0089] Perform partial autocorrelation processing on the target advertising data set to obtain a second correlation graph.

[0090] In this embodiment, after confirming that the time series is stationary and non-white noise, use the autocorrelation function (ACF, Autocorrelation Function) and the partial autocorrelation function (PACF, Partial Autocorrelation Function) graphs to identify the potential order (p, d, q) of the autoregressive integrated moving average model (ARIMA, AutoRegressive Integrated Moving Average) model;

[0091] where the autocorrelation function ACF(k) measures the correlation between the value of the time series at lag k and the current value:

[0092]

[0093] In the ACF graph, if the autocorrelation coefficient decays rapidly to zero after a certain lag period q, it is usually recommended to select the order q of the moving average term as this lag period;

[0094] The partial autocorrelation function PACF(k) measures the correlation between the value at lag k and the current value after controlling for the effects of all smaller lag periods:

[0095]

[0096] It represents that after controlling the effects of the intermediate lag terms y t-1 , y t-2 , …, y t-(k-1) , the direct correlation between the time series y t and its lag k - order version y t-k ;

[0097] Among them, Cov represents covariance; Var represents variance; the number of differencing times d is the number of differencing times required to make the time series stationary, which is usually determined by observing the trend and seasonality of the time series;

[0098] The basic form of the differencing operation is:

[0099] Δy t = y t - y t-1

[0100] Through the above steps, the appropriate order (p, d, q) of the ARIMA model is determined.

[0101] In an optional embodiment of the present invention, step 15 includes:

[0102] Step 151, input the first - layer processing result and part of the target advertisement dataset into the input layer of the second prediction layer of the advertisement spillover value prediction model to obtain an input - layer result;

[0103] Step 152, input the input - layer structure into the hidden layer of the second prediction layer of the advertisement spillover value prediction model to obtain a hidden - layer result;

[0104] Step 153, input the hidden - layer result into the output layer of the second prediction layer of the advertisement spillover value prediction model to obtain the target advertisement spillover value.

[0105] In this embodiment, the second prediction layer of the advertisement spillover value prediction model is trained according to a second preset network model, where the second preset network model can be an LSTM network (Long Short - Term Memory network);

[0106] Input the processing result of the first layer and part of the target advertisement dataset into the input layer of the second prediction layer of the advertisement spillover value prediction model to obtain the input layer result. Specifically, input the processing result of the first layer and part of the target advertisement dataset into the input layer of the trained LSTM network to obtain the input layer result;

[0107] Input the input layer structure into the hidden layer of the second prediction layer of the advertisement spillover value prediction model to obtain the hidden layer result. Specifically, input the input layer structure into the hidden layer of the trained LSTM network to obtain the hidden layer result;

[0108] Input the hidden layer result into the output layer of the second prediction layer of the advertisement spillover value prediction model to obtain the target advertisement spillover value. Specifically, input the hidden layer result into the output layer of the trained LSTM network to obtain the target advertisement spillover value;

[0109] In this embodiment, in order to further enhance the prediction ability, after completing the processing of the first prediction layer of the advertisement spillover value prediction model, input the processing result of the first layer and part of the target advertisement dataset into the trained LSTM network for prediction; specifically, adopt a time window with a length of t, and integrate the historical advertisement features and user features within this window into a multi-dimensional input vector; these features include but are not limited to the number of likes, forwards, comments of the advertisement, and the gender, age group, etc. of the user. Input these multi-dimensional input vectors as part of the target advertisement dataset and the processing result of the first layer into the trained LSTM network for prediction to obtain the target advertisement spillover value;

[0110] Enable the LSTM network to learn richer temporal patterns, so as to more accurately predict the data points at time t+1; in order to avoid the overfitting phenomenon of the LSTM model, take a series of regularization measures, apply the random dropout (Dropout) technique to randomly discard some neurons to reduce the network complexity and force the remaining neurons to learn more robust features;

[0111] Training process of the LSTM network model: Input the prepared time series data as the training set into the network. In the training stage, adopt the backpropagation through time (BPTT) algorithm, combined with the gradient descent method to iteratively update the network parameters θ * , to minimize the selected loss function, and select the mean squared error (MSE) as the standard to measure the difference between the predicted output and the true value:

[0112]

[0113] During the model validation process, closely monitor its performance on the independent validation set and adjust the hyperparameters accordingly, such as the learning rate, the number of network layers, and the number of units in each layer, etc.; after completing the training, comprehensively evaluate the model and record multiple performance indicators such as MAE, MSE, and coefficient of determination (R 2 , Coefficient of Determination) as the basis for comprehensive evaluation; among them, the formula for the coefficient of determination is as follows:

[0114]

[0115] where n represents the number of samples; y i represents the actual value; represents the predicted value;

[0116] After multiple rounds of iterative optimization, select the LSTM model with the best performance for actual application deployment;

[0117] During the training process, set the output of each neuron to 0 with a certain probability. L2 regularization is to add a penalty term to the loss function to limit the size of the model parameters, thereby reducing the model complexity and avoiding overfitting; for the LSTM network, the penalty term is usually the product of the sum of squares of all weights and a small constant:

[0118]

[0119] where L represents the original loss function; λ represents the regularization strength hyperparameter; w i represents the weight parameter in the model.

[0120] As Figure 2 shown, the prediction process of the advertising spillover value:

[0121] Obtain the front-end advertising dataset, where the front-end advertising dataset includes but is not limited to features such as the consumption, interaction, publisher, user, etc. of the advertisement, and the total spillover value generated by the advertisement every day;

[0122] Preprocess the front-end advertising dataset, perform a filling process on the front-end advertising dataset to obtain the first preprocessed dataset; perform an outlier process on the first preprocessed dataset to obtain the second preprocessed dataset; perform a standardization process on the second preprocessed dataset to obtain the transitional advertising dataset;

[0123] Perform inspection processing on the transitional advertising data set, perform unit root detection processing on the transitional advertising data set to obtain a first detection result; perform autocorrelation detection processing on the first detection result to obtain a target advertising data set; wherein, if the sequence is non-stationary, it is necessary to check the sequence, make the sequence stationary, and then perform autocorrelation detection; if the autocorrelation detection result is significant, perform subsequent steps; if the autocorrelation detection result is not significant, re-acquire the front-end advertising data set, and perform the above steps until the autocorrelation detection result is significant;

[0124] Determine the potential order of the ARIMA model, and determine the potential order (p, d, q) of the ARIMA model through ACF and PACF graphs;

[0125] Input the target advertisement data set into the trained ARIMA model for processing to obtain the first layer processing result;

[0126] Inputting the first layer processing result and part of the target advertisement data set into the trained LSTM network model to obtain the target advertisement spillover value;

[0127] Through the above process, the specific spillover value of the advertisement can be directly predicted, so that advertisers can adjust budget allocation and optimize delivery strategies more finely according to the specific spillover value to ensure effective use of resources.

[0128] like Figure 3 As shown, an embodiment of the present invention further provides a prediction device 30 for advertising overflow value, comprising:

[0129] An acquisition module 31 is used to acquire a front-end advertisement data set;

[0130] The processing module 32 is used to pre-process the front-end advertising data set to obtain a transition advertising data set; perform inspection processing on the transition advertising data set to obtain a target advertising data set; input the target advertising data set into the first prediction layer of the advertising spillover value prediction model for processing to obtain a first layer processing result; input the first layer processing result and part of the target advertising data set into the second prediction layer of the advertising spillover value prediction model to obtain a target advertising spillover value; wherein the first prediction layer of the advertising spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertising spillover value prediction model is trained according to a second preset network model.

[0131] Optionally, preprocessing the front-end advertisement data set to obtain a transition advertisement data set includes:

[0132] Performing gap filling processing on the front-end advertisement data set to obtain a first pre-processed data set;

[0133] Perform outlier processing on the first preprocessed data set to obtain a second preprocessed data set;

[0134] Perform standardization processing on the second preprocessed data set to obtain a transitional advertisement data set.

[0135] Optionally, perform inspection processing on the transitional advertisement data set to obtain a target advertisement data set, including:

[0136] Perform unit root test processing on the transitional advertisement data set to obtain a first test result;

[0137] Perform autocorrelation test processing on the first test result to obtain a target advertisement data set.

[0138] Optionally, input the target advertisement data set into the first prediction layer of an advertisement spillover value prediction model for processing to obtain a first-layer processing result, including:

[0139] Input the target advertisement data set into the first processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first processing result;

[0140] Input the first processing result into the second processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a second processing result;

[0141] Input the second processing result into the third processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first-layer processing result.

[0142] Optionally, it further includes:

[0143] Process the target advertisement data set to obtain an autocorrelation graph;

[0144] Determine the potential order of the first prediction layer of the advertisement spillover value prediction model according to the autocorrelation graph.

[0145] Optionally, process the target advertisement data set to obtain an autocorrelation graph, including:

[0146] Perform autocorrelation processing on the target advertisement data set to obtain a first correlation graph;

[0147] Perform partial autocorrelation processing on the target advertisement data set to obtain a second correlation graph.

[0148] Optionally, input the first-layer processing result and a part of the target advertisement data set into the second prediction layer of the advertisement spillover value prediction model to obtain a target advertisement spillover value, including:

[0149] Input the first layer processing result and a part of the target advertisement data set into the input layer of the second prediction layer of the advertisement spillover value prediction model to obtain an input layer result;

[0150] Input the input layer structure into the hidden layer of the second prediction layer of the advertisement spillover value prediction model to obtain a hidden layer result;

[0151] Input the hidden layer result into the output layer of the second prediction layer of the advertisement spillover value prediction model to obtain the target advertisement spillover value.

[0152] It should be noted that this device corresponds to the above method. All implementation manners in the above method embodiments are applicable to the embodiments of this device and can also achieve the same technical effects.

[0153] An embodiment of the present invention also provides a computing device, including: a processor and a memory storing a computer program. When the computer program is run by the processor, it executes the method as described in the above embodiments. All implementation manners in the above method embodiments are applicable to this embodiment and can also achieve the same technical effects.

[0154] An embodiment of the present invention also provides a computer-readable storage medium storing an instruction. When the instruction runs on a computer, it causes the computer to execute the method as described in the above embodiments. All implementation manners in the above method embodiments are applicable to this embodiment and can also achieve the same technical effects.

[0155] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraint conditions of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0156] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0157] In the embodiments provided by the present invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the apparatus or unit can be in an electrical, mechanical or other form.

[0158] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0159] In addition, in each embodiment of the present invention, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0160] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or part of this 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 for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0161] In addition, it should be noted that in the devices and methods of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations shall be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed in chronological order according to the described order, but it is not necessary to execute them in chronological order. Some steps can be executed in parallel or independently of each other. For those of ordinary skill in the art, it is understandable that all or any steps or components of the method and device of the present invention can be implemented in any computing device (including processors, storage media, etc.) or a network of computing devices in the form of hardware, firmware, software, or a combination thereof, which can be achieved by those of ordinary skill in the art using their basic programming skills after reading the description of the present invention.

[0162] Therefore, the object of the present invention can also be achieved by running a program or a set of programs on any computing device. The computing device can be a well-known general-purpose device. Therefore, the object of the present invention can also be achieved only by providing a program product containing program code for implementing the method or device. That is to say, such a program product also constitutes the present invention, and a storage medium storing such a program product also constitutes the present invention. Obviously, the storage medium can be any well-known storage medium or any storage medium developed in the future. It should also be noted that in the devices and methods of the present invention, obviously, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations shall be regarded as equivalent solutions of the present invention. Moreover, the steps of performing the above series of processes can naturally be executed in chronological order according to the described order, but it is not necessary to execute them in chronological order. Some steps can be executed in parallel or independently of each other.

[0163] The above is the preferred embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for predicting advertising spillover value, characterized in that: include: Get the front-end advertising dataset; Preprocessing the front-end advertisement data set to obtain a transition advertisement data set; Performing inspection processing on the transition advertisement data set to obtain a target advertisement data set; Inputting the target advertisement data set into the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first layer processing result; The first layer processing results and part of the target advertising data set are input into the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value; wherein the first prediction layer of the advertising spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertising spillover value prediction model is trained according to a second preset network model.

2. The method for predicting advertising spillover value according to claim 1, characterized in that: Preprocessing the front-end advertisement data set to obtain a transition advertisement data set includes: Performing gap filling processing on the front-end advertisement data set to obtain a first pre-processed data set; Performing outlier processing on the first preprocessed data set to obtain a second preprocessed data set; The second preprocessed data set is standardized to obtain a transition advertisement data set.

3. The method for predicting advertising spillover value according to claim 1, characterized in that: The transition advertisement data set is inspected to obtain a target advertisement data set, including: Performing unit root detection processing on the transition advertisement data set to obtain a first detection result; The first detection result is subjected to autocorrelation detection processing to obtain a target advertisement data set.

4. The method for predicting advertising spillover value according to claim 1, characterized in that: The target advertisement data set is input into the first prediction layer of the advertisement spillover value prediction model for processing to obtain the first layer processing result, including: Inputting the target advertisement data set into a first processing module of a first prediction layer of an advertisement spillover value prediction model for processing to obtain a first processing result; Inputting the first processing result into the second processing module of the first prediction layer of the advertising spillover value prediction model for processing to obtain a second processing result; The second processing result is input into the third processing module of the first prediction layer of the advertisement spillover value prediction model for processing to obtain the first layer processing result.

5. The method for predicting advertising spillover value according to claim 4, characterized in that: Also includes: Process the target advertising data set to obtain an autocorrelation graph; The potential order of the first prediction layer of the advertisement spillover value prediction model is determined according to the autocorrelation graph.

6. The method for predicting advertising spillover value according to claim 5, characterized in that: The target advertising data set is processed to obtain the autocorrelation graph, including: Performing autocorrelation processing on the target advertisement data set to obtain a first correlation graph; The target advertisement data set is subjected to partial autocorrelation processing to obtain a second correlation graph.

7. The method for predicting advertising spillover value according to claim 1, characterized in that: Inputting the first layer processing result and part of the target advertisement data set into the second prediction layer of the advertisement spillover value prediction model to obtain the target advertisement spillover value, including: Inputting the first layer processing result and part of the target advertisement data set into the input layer of the second prediction layer of the advertisement spillover value prediction model to obtain the input layer result; Inputting the input layer structure into the hidden layer of the second prediction layer of the advertising spillover value prediction model to obtain a hidden layer result; The hidden layer result is input into the output layer of the second prediction layer of the advertisement spillover value prediction model to obtain the target advertisement spillover value.

8. A device for predicting advertising spillover value, characterized in that: include: The acquisition module is used to obtain the front-end advertising data set; A processing module, used for preprocessing the front-end advertisement data set to obtain a transition advertisement data set; Performing inspection processing on the transition advertisement data set to obtain a target advertisement data set; Inputting the target advertisement data set into the first prediction layer of the advertisement spillover value prediction model for processing to obtain a first layer processing result; The first layer processing results and part of the target advertising data set are input into the second prediction layer of the advertising spillover value prediction model to obtain the target advertising spillover value; wherein the first prediction layer of the advertising spillover value prediction model is trained according to a first preset network model, and the second prediction layer of the advertising spillover value prediction model is trained according to a second preset network model.

9. A computing device, characterized in that include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is performed.

10. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images