Data processing method and device, equipment, storage medium and computer program product

Abstract

A data processing method, device, equipment, storage medium and computer program product, the method comprising: obtaining observation data of each observation object in a plurality of observation objects of a target application; wherein the observation data comprises N intervention events corresponding to each observation time in at least one observation time and the number of intervention events of each intervention event, and N is an integer greater than or equal to 0; determining the intervention effect parameter of each intervention event included in the observation data according to the observation data of each observation object and a target regression method; determining a target intervention event according to the intervention effect parameter of each intervention event, and executing the target intervention event on the object using the target application. By using the method of the present application, the granularity of intervention effect measurement can be refined, and the effectiveness of intervention strategy formulation can be improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a data processing method, a data processing apparatus, a computer device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] After implementing an intervention, it's typically required to evaluate its effectiveness. For example, after a game app launches a new activity, it's necessary to assess whether the activity helps increase player engagement. Currently, it's often assumed that only one intervention event exists throughout the observation period, and that the object is only affected by that event once. However, in real-world business scenarios, the object being intervened in may be affected by multiple different intervention events, and the same intervention event may be applied multiple times. Therefore, this assumption can obscure the effects of different intervention events and lead to incorrect evaluation results. Summary of the Invention

[0003] This application provides a data processing method, apparatus, device, storage medium, and computer program product that can improve the effectiveness of intervention strategy formulation by refining the granularity of intervention effect measurement.

[0004] On one hand, embodiments of this application provide a data processing method, the method comprising:

[0005] Obtain observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0;

[0006] Based on the observation data of each observation object and the target regression method, determine the intervention effect parameters of each intervention event included in the observation data;

[0007] The target intervention event is determined based on the intervention effect parameters of each intervention event, and the target intervention event is executed on the object using the target application.

[0008] On one hand, embodiments of this application provide a data processing apparatus, the apparatus comprising:

[0009] An acquisition unit is used to acquire observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0;

[0010] The processing unit is used to determine the intervention effect parameters of each intervention event included in the observation data based on the observation data of each observation object and the target regression method;

[0011] The processing unit is further configured to determine a target intervention event based on the intervention effect parameters of each intervention event, and to execute the target intervention event on the object using the target application.

[0012] On one hand, embodiments of this application provide a computer device, which includes a processor, a communication interface, and a memory. The processor, the communication interface, and the memory are interconnected. The memory stores a computer program, and the processor is used to call the computer program to execute the data processing method of any of the above possible implementations.

[0013] On one hand, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements a data processing method of any possible implementation.

[0014] Accordingly, this application also provides a computer program product, which includes a computer program or computer instructions, and the computer program or computer instructions are executed by a processor to implement the steps of the data processing method provided in this application.

[0015] Accordingly, this application also provides a computer program, which includes computer instructions stored in a computer-readable storage medium. The 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 provided in this application.

[0016] In this embodiment, observation data for each of the multiple observation objects of the target application can be obtained. The observation data includes N intervention events corresponding to each observation time within at least one observation time, and the number of interventions for each intervention event, where N is an integer greater than or equal to 0. Based on the observation data of each observation object and the target regression method, intervention effect parameters for each intervention event included in the observation data are determined. Target intervention events are determined based on the intervention effect parameters of each intervention event, and the target intervention events are executed on the objects using the target application. By employing this embodiment, the granularity of intervention effect measurement can be refined, and the intervention effects under different intervention events can be evaluated, thereby enabling the development of more effective intervention strategies for the objects using the target application. Attached Figure Description

[0017] To more clearly illustrate the technical methods of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram illustrating the intervention effect of a traditional double difference method provided in an embodiment of this application;

[0019] Figure 2 This application provides a schematic diagram of the system architecture of a data processing system.

[0020] Figure 3 A flowchart illustrating a data processing method provided in this application embodiment. Figure 1 ;

[0021] Figure 4 This application provides a schematic diagram of a pop-up interface for a friend recall activity.

[0022] Figure 5 A flowchart illustrating a data processing method provided in this application embodiment. Figure 2 ;

[0023] Figure 6 A schematic diagram of a data configuration interface provided in an embodiment of this application;

[0024] Figure 7 A flowchart illustrating a data processing method provided in this application embodiment. Figure 3 ;

[0025] Figure 8 This is a schematic diagram of the structure of a data processing device provided in an embodiment of this application;

[0026] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0027] The technical methods in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0028] The difference-in-differences (DID) method is often used to evaluate the effect of an intervention event and is a quasi-experimental solution for causal inference. Traditional DID schemes typically assume two groups of participants: an intervention group and a control group, and two intervention phases: before and after the intervention. Before the intervention, neither the intervention group nor the control group is affected by the intervention event; after the intervention, only the intervention group is affected by the intervention event. See also... Figure 1 The graph shows the time axis on the horizontal axis and the response data that measures the effectiveness of the intervention on the vertical axis. Figure 1 It can be seen that C at time t (after intervention) represents the response result data of the intervention group at time t, and E at time t represents the hypothetical response result data of the intervention group at time t without intervention. Therefore, the intervention effect ATT received by the intervention group can be determined as follows:

[0029] ATT = E[Y] 1t -Y 0t |G=1]=CE (1)

[0030] Among them, Y 1t Y represents the response outcome data of the intervention group after the intervention and while receiving the intervention. 0t This represents the response data of the intervention group after intervention but without intervention. G=1 represents the intervention group, and G=0 represents the control group.

[0031] But in real-world scenarios, Y 0t Since it is unobservable, the difference-in-differences method introduces a control group as the counterfactual state of the intervention group before intervention. The intervention effect (ATT) is obtained by comparing the changes in the intervention group and the control group before and after intervention.

[0032] ATT = E[Y] 1t -Y 1t-1 |G=1]-E(Y 0t -Y 0t-1 |G=0)]=(CA)-(DB) (2)

[0033] Among them, E[Y 1t -Y 1t-1 |G=1] represents the response outcome data of the intervention group after the intervention minus the response outcome data of the intervention group before the intervention; E(Y 0t -Y 0t-1 [G=0] represents the response data of the control group after the intervention minus the response data of the control group before the intervention.

[0034] In practical applications, the difference-in-differences method typically calculates the intervention effect ATT by constructing a linear regression equation. A standard difference-in-differences linear regression equation can be expressed as follows:

[0035] Y it =β1*Post t +β2*Intervention i +β3*Post t *Intervention i +ε (3)

[0036] Among them, Y it Y represents the response outcome data of subject i (including the intervention group and the control group) at intervention phase t (including before and after the intervention). Specifically, Y 11 Y represents the response outcome data of the intervention group after the intervention. 10 Y represents the response outcome data of the intervention group before the intervention. 01 Y represents the response data of the control group after intervention. 00 This represents the response data of the control group before the intervention; Post t The value of is {0, 1}, when Post t A value of 0 indicates the condition before intervention, when Post t A value of 1 indicates intervention; Intervention i The value of is {0, 1}, when Intervention i A value of 0 indicates the control group, while an intervention value of 0 indicates the control group. i A value of 1 indicates the intervention group; β1 represents the time effect, β1 = DB; β2 represents the inter-group difference, β2 = AB; β3 represents the intervention effect ATT, β3 = (CA) - (DB); ε represents the interference term.

[0037] Because Y 11 =β1+β2+β3+ε, Y 10 =β2+ε,Y 01 =β1+ε,Y 00 =ε, therefore according to equation (2), we can obtain (Y) 11 -Y 10 )-(Y 01 -Y 00 ) = β3, so β3 represents the intervention effect ATT.

[0038] Post t and Intervention i Y is the independent variable in this double-difference linear regression equation. itβ1, β2, and β3 are the dependent variables in the double difference linear regression equation, and β1, β2, and β3 are the regression parameters to be determined in the double difference linear regression equation. β1, β2, and β3 can be obtained by estimating the parameters of the double difference linear regression equation. Since β3 represents the intervention effect ATT, the intervention effect ATT is obtained when β3 is solved.

[0039] As explained above, the traditional difference-in-differences method is only applicable when there is only one intervention event throughout the entire observation process, and the number of interventions is a single frequency. However, in real-world scenarios, the object being intervened upon is usually affected by more than one intervention event, and the number of interventions is not limited to just one. Therefore, this application proposes an improved difference-in-differences method based on the traditional method, which can estimate the intervention effect under different intervention events and different numbers of interventions, expanding the applicability of the traditional difference-in-differences method and refining the granularity of intervention effect measurement. This improved difference-in-differences method is illustrated through the following examples.

[0040] Please see Figure 2 , Figure 2 This application provides a schematic diagram of the system architecture of a data processing system. Figure 2 The system architecture shown can be used to implement the data processing method proposed in the embodiments of this application. For example... Figure 2 As shown, the system architecture includes: server 20 and multiple terminal devices 21.

[0041] The server 20 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. The terminal device 21 can be a smartphone, tablet, laptop, desktop computer, intelligent voice interaction device, smart home appliance, in-vehicle terminal, etc., but is not limited to these. Figure 2 The terminal device 21 shown is connected to the server 20 via a network.

[0042] The terminal device 21 deploys a target application. A user using the target application can be considered an object using the target application. It should be noted that users are distinguished by their application accounts logged into the target application. User behavior data related to the target application can be sent from the terminal device 21 to the server 20. Examples include game match data and activities participated in by game players within the game application. In this case, the server 20 essentially acts as the application server corresponding to the target application. In one implementation, the server 20 can connect to the application server corresponding to the target application via a network to obtain relevant data about the object using the target application, after obtaining the user's permission, for evaluating the intervention effect.

[0043] In one embodiment, server 20 can identify multiple observation objects from all objects using the target application, obtain observation data for each observation object, and use the observation data of each observation object and a target regression method to determine the intervention effect parameters of each intervention event included in the observation data. Then, based on the intervention effect parameters of each intervention event, a target intervention event is determined, and the target intervention event is executed on the objects using the target application. The observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0. Therefore, based on the intervention effect parameters determined by the observation data and the target regression method, the intervention effect under different intervention events can be evaluated, thereby selecting the most effective intervention event as the target intervention event to intervene on the objects using the target application, which can improve the effectiveness of the intervention strategy formulation.

[0044] It is understood that the system architecture diagrams described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of system architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0045] The above provides a brief overview of the data processing method provided in the embodiments of this application. The specific implementation of this data processing method will be described in detail below.

[0046] Please see Figure 3 , Figure 3 A flowchart illustrating a data processing method provided in this application embodiment. Figure 1 This method is applied to the above. Figure 2 Server 20 in the method. The method includes the following steps S301-S303:

[0047] S301. Obtain observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, and N is an integer greater than or equal to 0.

[0048] The observational data refers to data collected through investigation or observation without any human control over the subject matter. The observation time is an observation window, the size of which can be a day, a year, etc., and this application does not limit this. For example, the at least one observation time could be December 11th, December 12th, December 13th, etc., or it could be 2019, 2020, 2021, etc. An intervention event refers to the content corresponding to an operation performed to achieve a certain purpose, such as algorithm iteration for a target application, or launching an activity in the target application.

[0049] The multiple observation objects can be some or all of the objects using the target application. These multiple observation objects include a first observation object and a second observation object. The first observation object is subject to at least one intervention event, and each intervention event can occur once or multiple times. The second observation object is not subject to any intervention event. That is, the first observation object is equivalent to the intervention group, and the second observation object is equivalent to the control group. For the second observation object, there are 0 intervention events at each observation time, and the number of interventions is also 0.

[0050] It should be noted that, in order to evaluate the effectiveness of one or more intervention events, the intervention needs to be implemented only on a subset of users of the target application. For example, such as... Figure 4The image shows the pop-up interface for a friend recall event in a shooting game. This shooting game can conduct friend recall events three times, on December 11th, December 18th, and December 25th. Each time, the pop-up interface can be pushed to only a subset of players in the game. Each time a player receives this pop-up, it's considered that the player has received one instance of intervention. This situation constitutes one or more interventions for the same event. Players who receive the pop-up can be considered the first observation group, and players who do not receive the pop-up can be considered the second observation group. Suppose that on December 11th, Player 1 was shown three pop-up windows for the friend recall event. The event was canceled between December 12th and 17th. Then, on December 18th, Player 1 was shown two more pop-up windows for the friend recall event. For Player 1, the observed data could include: Player 1's intervention on December 11th corresponding to intervention event 1 (the friend recall event pop-up), with intervention event 1 occurring 3 times; Player 1's intervention on December 12th-17th corresponding to intervention event 1 (the friend recall event pop-up), with intervention event 1 occurring 3 times; and Player 1's intervention on December 18th corresponding to intervention event 1 (the friend recall event pop-up), with intervention event 1 occurring 5 times. That is, the number of interventions for each intervention event is obtained by summing them up.

[0051] For example, to boost user activity in a game app, Activities 1, 2, and 3 were launched from November 4-19, November 19-30, and December 2-10, respectively. Similarly, each activity could be pushed to only a subset of players within the game app, and each activity could be pushed once or multiple times. This scenario involves multiple intervention events intervening once or multiple times. Players who received at least one activity can be considered the first observation group, and players who did not receive any activity can be considered the second observation group. Assuming that Activity 1 is pushed to Player 1 once a day from November 4th to 19th, and Activity 2 is pushed to Player 1 once a day from November 19th to 30th, then for Player 1, the observation data could include: Player 1 corresponds to Intervention Event 1 (Activity 1) on November 4th, and the number of interventions for Intervention Event 1 is 1; Player 1 corresponds to Intervention Event 1 (Activity 1) on November 5th, and the number of interventions for Intervention Event 1 is 2; and so on up to November 19th, Player 1 corresponds to Intervention Event 1 (Activity 1) on November 9th and the number of interventions for Intervention Event 2 is 2. Event 2 (Activity 2), and the number of interventions for Event 1 is 16, while the number of interventions for Event 2 is 1; Player 1 corresponds to Event 1 (Activity 1) and Event 2 (Activity 2) on November 10, and the number of interventions for Event 1 is 16, while the number of interventions for Event 2 is 2, and so on until November 30, Player 1 corresponds to Event 1 (Activity 1) and Event 2 (Activity 2) on November 30, and the number of interventions for Event 1 is 16, while the number of interventions for Event 2 is 12.

[0052] S302. Based on the observation data of each observation object and the target regression method, determine the intervention effect parameters of each intervention event included in the observation data.

[0053] The observation data includes each intervention event, referring to all intervention events that occurred during the observation process. For example, Player 1's observation data includes intervention event 1 on November 12th and intervention event 2 on November 13th, while Player 2's observation data includes intervention event 1 and intervention event 3 on November 12th. Therefore, the observation data includes all intervention events: intervention event 1, intervention event 2, and intervention event 3. The target regression method refers to regression using the improved difference-in-differences method provided in this application. This target regression method can obtain intervention effect parameters for each intervention event included in the observation data based on the observation data of each observation object. These intervention effect parameters can evaluate the relationship between each intervention event included in the observation data and the intervention effect. Furthermore, it can also evaluate the relationship between the number of interventions and the intervention effect.

[0054] S303. Determine the target intervention event based on the intervention effect parameters of each intervention event, and execute the target intervention event on the object using the target application.

[0055] The larger the intervention effect parameter, the better the intervention effect of the intervention event. In one implementation, the intervention event corresponding to the maximum intervention effect parameter can be used as the target intervention event, or the intervention event corresponding to the intervention effect parameter greater than the preset effect parameter can be used as the target intervention event, and then the target intervention event can be executed on all or some objects using the target application. For example, if the target intervention event is Activity 1 in Activity 1, Activity 2, and Activity 3, then Activity 1 will be pushed to all or some objects using the target application.

[0056] In this embodiment, observation data for each of the multiple observation objects of the target application can be obtained. The observation data includes N intervention events corresponding to each observation time within at least one observation time, and the number of interventions for each intervention event, where N is an integer greater than or equal to 0. Based on the observation data of each observation object and the target regression method, intervention effect parameters for each intervention event included in the observation data are determined. Target intervention events are determined based on the intervention effect parameters of each intervention event, and the target intervention events are executed on the objects using the target application. By employing this embodiment, the granularity of intervention effect measurement can be refined, and the intervention effects under different intervention events can be evaluated, thereby enabling the development of more effective intervention strategies for the objects using the target application.

[0057] Please see Figure 5 , Figure 5 A flowchart illustrating a data processing method provided in this application embodiment. Figure 2 This method is applied to the above. Figure 2 Server 20 in the method. The method includes the following steps S501-S505:

[0058] S501. Obtain observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, and N is an integer greater than or equal to 0.

[0059] Mixed cross-sectional data refers to data obtained by randomly sampling a population at different time points. Furthermore, the sampling at each time point is independent, resulting in different samples each time. Panel data, on the other hand, refers to data obtained by continuously observing the same population at different time points. In the embodiments of this application, the processing methods for mixed cross-sectional data and panel data are different.

[0060] Regarding panel data:

[0061] In one embodiment, acquiring observation data for each of a plurality of observation objects in a target application includes: determining a plurality of observation times and a plurality of observation objects in the target application. The plurality of observation objects in the target application may be a subset of objects determined from all objects using the target application, wherein the plurality of observation objects includes a first observation object and a second observation object.

[0062] It should be noted that, because it is necessary to examine two intervention phases—before and after—the intervention time for each intervention event included in the observation data is within these multiple observation periods. For example, if Activity 1, Activity 2, and Activity 3 were launched from November 4-19, November 19-30, and December 2-10 respectively, then these multiple observation periods can include each day from November 1 to December 30.

[0063] Furthermore, for each of the multiple observation objects, the intervention data, response result data, and dynamic feature data corresponding to each observation time in the multiple observation times are obtained.

[0064] The intervention data for each observation time includes N intervention events and the number of interventions for each event. For example, if Player 1 was shown a friend recall activity pop-up three times on December 11th, then the intervention data for Player 1 on December 11th includes: Player 1 was shown intervention event 1 (the friend recall activity pop-up) on December 11th, and the number of interventions for intervention event 1 was 3. Response result data is used to measure the intervention effect, such as the number of times players participated in the game, weekly retention rate, bi-weekly retention rate, monthly retention rate, etc. Users who start using the application within a certain period and continue using it after a period of time are considered retained; the proportion of these users to the total number of users at that time is the retention rate. Dynamic feature data refers to the variable individual characteristic data of the observed object, such as the observed object's online time, online duration, and current expenditure on December 11th. It should be noted that dynamic feature data is obtained after obtaining the observed object's permission.

[0065] Finally, the intervention data, response result data, and dynamic feature data corresponding to each observation object at each observation time are used as the observation data for each observation object. It can be seen that, in this way, the intervention data, response result data, and dynamic feature data corresponding to each observation object at each observation time across multiple observation objects can be obtained, thus acquiring panel data.

[0066] For mixed cross-sectional data:

[0067] In one embodiment, obtaining observation data for each of the multiple observation objects of the target application includes: determining M observation objects corresponding to each observation time in the multiple observation times of the target application, where M is an integer greater than 0. That is, for each observation time, a portion of objects can be extracted from all objects using the target application as observation objects.

[0068] Furthermore, for each of the M observed objects, static feature data is obtained. Static feature data consists of immutable individual characteristics of the observed object, such as its age and gender. It should be noted that this static feature data is obtained after obtaining permission from the observed object.

[0069] Furthermore, intervention data, response result data, and dynamic characteristic data for each observed object at the corresponding observation time are obtained. If Player 1 was selected as the observed object on December 11, then December 11 is the corresponding observation time for Player 1.

[0070] The intervention data corresponding to the observation time includes N intervention events corresponding to the observation time and the number of interventions for each intervention event. For example, if Player 1 was shown a friend recall activity pop-up three times on December 11th, then the intervention data for Player 1 on December 11th includes: Player 1 was shown intervention event 1 (the friend recall activity pop-up) on December 11th, and the number of interventions for intervention event 1 was 3. Response result data is used to measure the intervention effect, such as the number of times players participated in the game, weekly retention rate, bi-weekly retention rate, monthly retention rate, etc. Dynamic feature data refers to the variable individual characteristic data of the observed object, such as the observed object's online time, online duration, current expenditure, etc. on December 11th. It should be noted that this dynamic feature data is obtained after obtaining the observed object's permission.

[0071] Finally, the static feature data of each observation object, as well as the intervention data, response result data, and dynamic feature data corresponding to each observation object at the corresponding observation time, are used as the observation data for each observation object. It can be seen that, in this way, M observation objects are extracted for different observation times, thus obtaining mixed cross-sectional data.

[0072] It should be noted that, before intervention, the response data of the first and second observation subjects need to show parallel trends over time (i.e., the same trend), which satisfies the parallel trend assumption.

[0073] S502. Obtain the target regression equation, which is determined based on the independent variables of the intervention event and the number of interventions.

[0074] For panel data, the target regression equation is the first regression equation, which is shown in equation (4) below:

[0075]

[0076] Among them, Times itk The number of interventions is the independent variable indicating the number of interventions for observation subject i at observation time t corresponding to intervention event k. ik The intervention event independent variable indicates whether the observed object i corresponds to an intervention event k during the observation process; Post itk γ represents the intervention time independent variable, which indicates whether the observation time t is before or after the intervention event k; t The independent variable representing the time effect corresponding to the observation time t, ε represents the dynamic effect independent variable of the observed object i at observation time t. it Let represent the interference term corresponding to the observed object i at observation time t, and K represent the number of all intervention events that occur during the observation process. It should be noted that the interference term is a random error term that can satisfy the Gaussian assumption and follow a normal distribution. It is used to transform the description of the expected result from a precise statement into a probabilistic description.

[0077] For ease of subsequent understanding, the parameters itk The subscripts i, t, and k are used for explanation. k represents the intervention event that occurs during the observation process. For example, k=1 represents activity 1, k=2 represents activity 2, and k=3 represents activity 3. i=1 represents the first observation object, and i=0 represents the second observation object. t=0 indicates that the observation time is before the intervention event k, and t=1 indicates that the observation time is after the intervention event k.

[0078] The first regression equation is a multiple linear regression equation proposed in this application for panel data based on the improved double difference method, where θ 1k (k=1,...,K), θ 2k (k=1,...,K) are the undetermined regression parameters in the first regression equation.

[0079] For mixed cross-sectional data, the target regression equation is the second regression equation, which is shown in equation (5) below:

[0080]

[0081] Among them, Times itkThe number of interventions is the independent variable indicating the number of interventions for observation subject i at observation time t corresponding to intervention event k. ik The intervention event independent variable indicates whether the observed object i corresponds to an intervention event k during the observation process; Post itk γ represents the intervention time independent variable, which indicates whether the observation time t is before or after the intervention event k; t The independent variable representing the time effect corresponding to the observation time t, μ represents the dynamic effect independent variable of observed object i at observation time t. i ε represents the static effect independent variable of observed object i. it Let represent the interference term corresponding to the observed object i at observation time t, and K represent the number of all intervention events that occur during the observation process. It should be noted that the interference term is a random error term that can satisfy the Gaussian assumption and follow a normal distribution. It is used to transform the description of the expected result from a precise statement into a probabilistic description.

[0082] The second regression equation is a multiple linear regression equation proposed in this application based on the improved double difference method, specifically for mixed cross-sectional data, where β 1k (k=1,...,K), θ 1k (k=1,...,K), θ 2k (k=1,...,K) are the undetermined regression parameters in the second regression equation.

[0083] S503. Determine the target matrix equation based on the observation data of each observation object and the target regression equation, and determine at least one regression parameter in the target matrix equation based on the independent variable matrix and dependent variable matrix included in the target matrix equation.

[0084] Regarding panel data:

[0085] In one embodiment, the parameter values ​​corresponding to each variable in the target regression equation are determined based on the observation data of each observation object to obtain the auxiliary regression equation corresponding to each observation object. This includes: determining the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the number of interventions independent variable in the first regression equation based on the N intervention events corresponding to each observation time in multiple observation times and the intervention data corresponding to each observation object in the target observation time. The target observation time is one of the multiple observation times.

[0086] Specifically, the parameter values ​​corresponding to the independent variables of the intervention events in the first regression equation are determined based on the N intervention events corresponding to each observation time across multiple observation periods for each observed object. For example, if K=3, activity 1 is represented by k=1, activity 2 by k=2, and activity 3 by k=3, then based on the N intervention events corresponding to player 1 and player 2 at each observation time, it can be seen that activities 1 and 3 were pushed to player 1 (who belongs to the first observed object). 11 =1, Intervention 12 =0, Intervention 13 =1, Activity 1 was pushed to player 2 (who is the first observed object), then Intervention 11 =1, Intervention 12 =0, Intervention 13 =0. That is, Intervention ik When = 1, it indicates that the observed object i corresponds to an intervention event k during the observation process. ik = 0 indicates that the observed object i did not correspond to any intervention event k during the observation process. The parameter value corresponding to the intervention frequency independent variable in the first regression equation is determined based on the intervention data corresponding to each observed object at the target observation time. That is, the parameter value corresponding to the intervention frequency independent variable is determined based on the N intervention events corresponding to each observed object at the target observation time, and the intervention frequency of each intervention event. For example, the start time of Activity 1 is December 9th, the start time of Activity 2 is December 11th, and the start time of Activity 3 is December 13th. Player 1 (belonging to the first observed object) corresponds to both Activity 1 and Activity 2 on December 11th, and the intervention frequency for Activity 1 is 3, while the intervention frequency for Activity 2 is 2. Therefore, Times... 111 =3, Times 112 =2, Times 103 =0. For example, if Player 1 (the first observed object) has 0 intervention events on December 8th, then Times... 101 =0,Times 102 =0,Times 103 =0. That is, Times itk This equals the number of interventions for intervention event k corresponding to observation time t for intervention subject i. It should be noted that the intervention for the second observation subject... ik Both are 0, and Times itk All are 0.

[0087] Furthermore, based on the intervention start time and multiple observation times of each intervention event included in the observation data, the target observation time is labeled to obtain the parameter values ​​corresponding to the intervention time independent variable and the parameter values ​​corresponding to the time effect independent variable in the first regression equation.

[0088] Specifically, the target observation time is labeled based on the intervention start time of each intervention event included in the observation data, resulting in the parameter values ​​corresponding to the intervention time independent variable in the first regression equation. For example, if Activity 1 starts on December 9th, Activity 2 starts on December 11th, and Activity 3 starts on December 13th, and Player 1 corresponds to both Activity 1 and Activity 2 on December 11th, then Post... 111 =1, Post 112 =1, Post 103 =0, meaning that for Activities 1 and 2 it is after the intervention, and for Activity 3 it is before the intervention. The target observation time is labeled based on multiple observation times to obtain the parameter values ​​corresponding to the time effect independent variables in the first regression equation. For example, if multiple observation times are each day from December 1st to 30th, and the target observation time is December 11th, then the element representing December 11th in the 1×30-dimensional vector corresponds to 1, and the other elements correspond to 0.

[0089] Furthermore, the dynamic characteristic data of each observed object at the target observation time are labeled to obtain the parameter values ​​corresponding to the dynamic effect independent variables in the first regression equation. For example, the parameter values ​​corresponding to the dynamic effect independent variables are [1, 0, 0, 0] for game duration of 0-6 hours, [0, 1, 0, 0] for game duration of 6-12 hours, [0, 1, 0, 0] for game duration of 12-18 hours, [0, 0, 1, 0] for game duration of 18-24 hours, and [0, 0, 0, 1]. Alternatively, 1, 2, 3, and 4 can be used to represent the parameter values ​​corresponding to the dynamic effect independent variables for game durations of 0-6 hours, 6-12 hours, 12-18 hours, and 18-24 hours, respectively. The response result data of each observed object at the target observation time is then used as the parameter value corresponding to the dependent variable in the first regression equation. For example, the number of times the game was played on December 9th is used as the parameter value corresponding to the dependent variable in the first regression equation. Finally, the auxiliary regression equation for each observed object was obtained.

[0090] Furthermore, the target matrix equation is determined using the auxiliary regression equation corresponding to each observation object. The independent variable matrix of the target matrix equation is determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation for each observation object, and the dependent variable matrix of the target matrix equation is determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation for each observation object.

[0091] To illustrate with an example, suppose K equals 3, and the auxiliary regression equations for each observation include:

[0092] Y1=θ 11 *0+θ 12 *0+θ 13 *1+θ 21 *0+θ 22 *0+θ 23 *3+4;

[0093] Y2=θ 11 *1+θ 12 *0+θ 13 *1+θ 21 *1+θ 22 *0+θ 23 *3+2;

[0094] Y3=θ 11 *1+θ 12 *0+θ 13 *0+θ 21 *4+θ 22 *0+θ 23 *0+3.

[0095] Then the independent variable matrix Dependent variable matrix Target matrix equation because so Then, based on the independent variable matrix and the dependent variable matrix, at least one regression parameter θ in the objective matrix equation can be determined. 1k (k=1,...,K), θ 2k (k = 1, ..., K), specifically θ here. 11 θ 12 θ 13 θ 21 θ 22 θ 23 .

[0096] For mixed cross-sectional data:

[0097] In one embodiment, the parameter values ​​corresponding to each variable in the target regression equation are determined based on the observation data of each observation object to obtain the auxiliary regression equation corresponding to each observation object. This includes: determining the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the number of interventions independent variable in the second regression equation based on the N intervention events corresponding to each corresponding observation time and the intervention data corresponding to each observation object at the target observation time. The target observation time is one of the multiple observation times.

[0098] Specifically, the parameter values ​​corresponding to the independent variables of the intervention events in the second regression equation are determined based on the N intervention events corresponding to each observation object at each corresponding observation time. For example, K=3, activity 1 is represented by k=1, activity 2 by k=2, and activity 3 by k=3. Based on the N intervention events corresponding to player 1 and player 2 at each corresponding observation time, it can be seen that activities 1 and 3 were pushed to player 1 (belonging to the first observation object). 11 =1, Intervention 12 =0, Intervention 13 =1, Activity 1 was pushed to player 2 (who is the first observed object), then Intervention 11 =1, Intervention 12 =0, Intervention 13 =0. That is, Intervention ik When = 1, it indicates that the observed object i corresponds to an intervention event k during the observation process. ik = 0 indicates that the observed object i did not correspond to any intervention event k during the observation process. The parameter value corresponding to the intervention frequency independent variable in the second regression equation is determined based on the intervention data corresponding to each observed object at the target observation time. That is, the parameter value corresponding to the intervention frequency independent variable is determined based on the N intervention events corresponding to each observed object at the target observation time, and the intervention frequency of each intervention event. For example, the start time of Activity 1 is December 9th, the start time of Activity 2 is December 11th, and the start time of Activity 3 is December 13th. Player 1 (belonging to the first observed object) corresponds to both Activity 1 and Activity 2 on December 11th, and the intervention frequency for Activity 1 is 3, while the intervention frequency for Activity 2 is 2. Therefore, Times... 111 =3, Times 112 =2, Times 103 =0. For example, if Player 1 (the first observed object) has 0 intervention events on December 8th, then Times... 101 =0,Times 102 =0,Times103 =0. That is, Times itk This equals the number of interventions for intervention event k corresponding to observation time t for intervention subject i. It should be noted that the intervention for the second observation subject... ik Both are 0, and Times itk All are 0.

[0099] Furthermore, based on the intervention start time and multiple observation times of each intervention event included in the observation data, the target observation time is labeled to obtain the parameter values ​​corresponding to the intervention time independent variable and the parameter values ​​corresponding to the time effect independent variable in the second regression equation.

[0100] Specifically, the target observation time is labeled according to the intervention start time of each intervention event included in the observation data, resulting in the parameter values ​​corresponding to the intervention time independent variable in the second regression equation. For example, if Activity 1 starts on December 9th, Activity 2 starts on December 11th, and Activity 3 starts on December 13th, and Player 1 corresponds to both Activity 1 and Activity 2 on December 11th, then Post... 111 =1, Post 112 =1, Post 103 =0, meaning that for Activities 1 and 2 it is after the intervention, and for Activity 3 it is before the intervention. The target observation time is labeled based on multiple observation times to obtain the parameter values ​​corresponding to the time effect independent variable in the second regression equation. For example, if multiple observation times are each day from December 1st to 30th, and the target observation time is December 11th, then the element representing December 11th in the 1×30-dimensional vector corresponds to 1, and the other elements correspond to 0.

[0101] Furthermore, the static feature data of each observed object and the dynamic feature data of each observed object at the target observation time are labeled to obtain the parameter values ​​corresponding to the static effect independent variables and the parameter values ​​corresponding to the dynamic effect independent variables in the second regression equation.

[0102] Specifically, the static feature data of each observed object at the target observation time are labeled to obtain the parameter values ​​corresponding to the static effect independent variables in the second regression equation. For example, if the gender is male, the parameter value corresponding to the static effect independent variable is 1; if the gender is female, the parameter value corresponding to the static effect independent variable is 0. The dynamic feature data of each observed object at the target observation time are labeled to obtain the parameter values ​​corresponding to the dynamic effect independent variables in the second regression equation. For example, if the game duration is 0-6 hours, the parameter value corresponding to the dynamic effect independent variable is [1, 0, 0, 0]; if the game duration is 6-12 hours, the parameter value corresponding to the dynamic effect independent variable is [0, 1, 0, 0]; if the game duration is 12-18 hours, the parameter value corresponding to the dynamic effect independent variable is [0, 0, 1, 0]; and if the game duration is 18-24 hours, the parameter value corresponding to the dynamic effect independent variable is [0, 0, 0, 1]. Alternatively, we can use 1, 2, 3, and 4 to represent the parameter values ​​corresponding to the dynamic effect independent variables for game durations of 0-6 hours, 6-12 hours, 12-18 hours, and 18-24 hours, respectively. Then, we use the response data of each observation object at the target observation time as the parameter value corresponding to the dependent variable in the second regression equation. For example, the number of times the observation object participated in the game on December 9th is used as the parameter value corresponding to the dependent variable in the second regression equation. Finally, we obtain the auxiliary regression equation for each observation object.

[0103] Furthermore, the target matrix equation is determined using the auxiliary regression equation corresponding to each observation object. The independent variable matrix of the target matrix equation is determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation for each observation object, and the dependent variable matrix is ​​determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation for each observation object. Then, based on the independent and dependent variable matrices, at least one regression parameter β in the target matrix equation can be determined. 1k (k=1,...,K), θ 1k (k=1,...,K), θ 2k (k = 1, ..., K).

[0104] In a feasible implementation, after obtaining at least one regression parameter in the target matrix equation, a robustness test, such as a placebo test, can be performed. After the robustness test passes, proceed to step S504 and subsequent steps.

[0105] S504. Determine the intervention effect parameters for each intervention event based on the at least one regression parameter.

[0106] At least one regression parameter determined for panel data includes θ 1k (k=1,...,K), θ 2k(k=1,...,K), this application will θ 1k (k=1,...,K), θ 2k (k=1,...,K) serve as the intervention effect parameters for each intervention event, where θ 1k This represents the second regression parameter, θ, which is included in the intervention effect parameters of intervention event k. 2k This indicates that the intervention effect parameters of intervention event k include the first regression parameter.

[0107] At least one regression parameter determined for mixed cross-sectional data includes β. 1k (k=1,...,K), θ 1k (k=1,...,K), θ 2k (k=1,...,K), this application will also include θ 1k (k=1,...,K), θ 2k (k=1,...,K) serve as the intervention effect parameters for each intervention event, where θ 1k This represents the second regression parameter, θ, which is included in the intervention effect parameters of intervention event k. 2k This indicates that the intervention effect parameters of intervention event k include the first regression parameter.

[0108] It should be noted that the intervention effect parameters include θ 1k (k = 1, ..., K) and θ 2k (k = 1, ..., K) is obtained through (Y) 11 -Y 10 )-(Y 01 -Y 00 It is determined. Assuming K is 1, that is, for a single intervention event, then according to equation (4), we can obtain: Y 11 =θ 11 +θ 21 *times+ψ1,Y 10 =ψ1,Y 01 =ψ1,Y 00 =ψ1, where ψ1 represents a value determined by time-effect independent variables, dynamic-effect independent variables, etc., because (Y 11 -Y 10 )-(Y 01 -Y 00 )=θ 11 +θ 21 *times, while θ 11 and θ 21 The larger the value, the better the intervention effect of the intervention event. According to equation (5), we can obtain: Y 11 =β 11 +θ 11 +θ 21 *times+ψ2,Y10 =β 11 +ψ2,Y 01 =ψ2,Y 00 =ψ2, and similarly, it will not be elaborated further.

[0109] S505. Determine the target intervention event based on the intervention effect parameters of each intervention event, and execute the target intervention event on the object using the target application.

[0110] In one embodiment, determining the target intervention event based on the intervention effect parameters of each intervention event includes: determining the effectiveness evaluation parameters of each intervention event based on the first regression parameter and the second regression parameter included in the intervention effect parameters of each intervention event. Specifically, the sum of the first regression parameter and the second regression parameter included in the intervention effect parameters of each intervention event is used as the effectiveness evaluation parameter of each intervention event. That is, θ 1k +θ 2k As a parameter for evaluating the effectiveness of intervention event k, for example, θ 11 +θ 21 As a parameter for evaluating the effectiveness of intervention event 1, θ 12 +θ 22 This serves as the effectiveness evaluation parameter for intervention event 2. Further, based on the effectiveness evaluation parameter, a target intervention event is determined from among the various intervention events. This can be either the intervention event corresponding to the maximum effectiveness evaluation parameter, or the intervention event corresponding to an effectiveness evaluation parameter greater than a preset evaluation parameter. The target intervention event is then executed on the objects using the target application.

[0111] In addition, the first regression parameter θ 2k It can be used to evaluate the relationship between the number of interventions and the intervention effect for each intervention event. For example, the first regression parameter θ 21 When θ is a positive number, 11 +θ 21 The larger the 'times' parameter is in the 'times' parameter, the better the intervention effect. The first regression parameter θ... 21 When θ is negative, 11 +θ 21The larger the *times* value, the worse the intervention effect. In one implementation, executing a target intervention event on objects using the target application includes: determining the correlation between the target intervention event and the number of interventions based on a first regression parameter included in the intervention effect parameters of the target intervention event. When the first regression parameter is positive, the correlation is positive; when the first regression parameter is negative, the correlation is negative. Further, the target number of interventions is determined based on the correlation. For example, if the correlation is positive, the target number of interventions can be higher, specifically determined from a first set, which includes multiple interventions greater than a preset number; if the correlation is negative, the target number of interventions can be lower, specifically determined from a second set, which includes multiple interventions less than or equal to a preset number. Finally, the target intervention event can be executed on objects using the target application according to the target number of interventions. For example, if Activity 1 is the target intervention event and the target number of interventions is 3, then Activity 1 can be pushed to objects using the target application 3 times.

[0112] In feasible embodiments, a target number of interventions can be determined manually first, and then based on θ... 1k +θ 2k *Target number of interventions: Determine the intervention effect of intervention event k at the target number of interventions, obtain the intervention effect of each intervention event at the target number of interventions, take the intervention event corresponding to the best intervention effect as the target intervention event, and execute the target intervention event on the object using the target application according to the target number of interventions.

[0113] θ 1k +θ 2k *times can be used to evaluate the intervention effect of intervention event k after *times* interventions. In one implementation, it can be based on θ 1k +θ 2k *times determines the intervention effect of intervention event k at each reference intervention number among multiple reference intervention numbers, obtains the intervention effect of each intervention event at each reference intervention number, takes the reference intervention number and intervention event corresponding to the best intervention effect as the target intervention number and target intervention event respectively, and executes the target intervention event on the object using the target application according to the target intervention number.

[0114] In this embodiment, the traditional double difference method is improved so that the improved double difference method can allow the observed object to be affected by multiple intervention events at the same time, which is more in line with the actual situation of intervention. Therefore, compared with the traditional double difference method, this solution has a wider range of applications and more realistic assumptions. At the same time, by evaluating the intervention effect under different intervention events and different number of interventions, better intervention strategies can be formulated to intervene in the objects using the target application.

[0115] In one embodiment, such as Figure 6 As shown, a data configuration interface is provided. The "Application Filter" option is used to determine the target application. The "Object ID" option is used to determine multiple observation objects. The "Analysis Period" option is used to determine multiple observation periods, for example, each day from December 3rd to December 31st is an observation period. The "Intervention Item Filter" option is used to determine each intervention event that occurs during the observation process; the intervention start time refers to the start time corresponding to each intervention event, for example, the intervention start time for intervention 1 is December 11th, 21st. The "Confusion Factor" option is used to determine the variable and immutable individual characteristic data of the observation objects. The "Outcome Indicator Selection" and "Business Indicator Selection" options are used to determine the response outcome data to measure the intervention effect; one or more indicators can be selected as the response outcome data. The "Regression Algorithm Selection" option is used to select a regression algorithm to estimate the undetermined regression parameters in the first or second regression equation. The least squares method finds the function that best fits the observation data by minimizing the sum of squared residuals.

[0116] See Figure 7 , Figure 7 A flowchart illustrating a data processing method provided in this application embodiment. Figure 3 This method is applied to the above. Figure 2 Server 20 in the middle. The method includes S701-S708:

[0117] S701, Obtain data. This can be based on... Figure 6 The provided data configuration interface allows you to determine multiple observation objects, multiple observation times, confounding factors, and response result data for measuring the intervention effect. Based on these multiple observation objects, multiple observation times, confounding factors, and response result data for measuring the intervention effect, you can obtain the observation data for each of the multiple observation objects.

[0118] S702. Constructing the dataset. Specifically, panel data or mixed cross-sectional data can be constructed using the observation data of each of the extracted observation objects.

[0119] S703. Labeling Processing. When performing labeling processing, it is necessary to clearly define the intervention group (i.e., the first observation subject), the control group (i.e., the second observation subject), and the intervention start time for each intervention event in order to perform labeling processing. For detailed implementation process, please refer to the aforementioned S503; this embodiment will not repeat it further.

[0120] S704. Hypothesis Testing. Specifically, this involves verifying the parallel trend hypothesis, which states that before the intervention, the response data of the intervention group and the control group should show parallel trends over time (i.e., the same trend).

[0121] S705. Model Construction. Construct a linear regression model based on the dataset type (including panel data or mixed cross-sectional data). Specifically, if the dataset type is panel data, the multiple auxiliary regression equations obtained according to equation (4) above constitute the constructed linear regression model; if the dataset type is mixed cross-sectional data, the multiple auxiliary regression equations obtained according to equation (5) above constitute the constructed linear regression model.

[0122] S706. Regression Parameter Estimation. The least squares method can be used to estimate the regression parameters in a linear regression model.

[0123] S707. Intervention Effect Evaluation. Specifically, the intervention effect of each intervention event can be quantified based on regression parameters. For detailed implementation process, please refer to the aforementioned S505; it will not be repeated in this embodiment.

[0124] S708. Robustness testing. This can specifically be a placebo test to validate the conclusions of the proposed solution and enhance their credibility. The approach to placebo testing is to use additional empirical evidence.

[0125] In feasible embodiments, multiple observation subjects can be classified according to the type of intervention event, the number of interventions, and the duration of intervention. After classification, the observation subjects included in each category are used as intervention groups, and linear regression models are established between each group and the control group.

[0126] In this embodiment, mixed cross-sectional data or panel data can be constructed by refining the individual granularity. At the same time, the regression equation in the traditional difference method is improved accordingly. For various types and frequencies of intervention events, the relationship between various intervention events and intervention times and intervention effects can be accurately measured.

[0127] It is understood that in the specific implementation of this application, individual characteristic data and other related data 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.

[0128] The methods of the embodiments of this application have been described in detail above. To facilitate better implementation of the methods of the embodiments of this application, the apparatus of the embodiments of this application is provided below. Please refer to... Figure 8 , Figure 8 This is a schematic diagram of the structure of a data processing device provided in an embodiment of this application. The data processing device 80 may include:

[0129] The acquisition unit 801 is used to acquire the observation data of each observation object among multiple observation objects of the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, and N is an integer greater than or equal to 0;

[0130] Processing unit 802 is used to determine the intervention effect parameters of each intervention event included in the observation data based on the observation data of each observation object and the target regression method;

[0131] The processing unit 802 is further configured to determine a target intervention event based on the intervention effect parameters of each intervention event, and to execute the target intervention event on the object using the target application.

[0132] In one embodiment, the acquisition unit 801 is specifically used to: acquire a target regression equation, wherein the target regression equation is determined based on the intervention event independent variable and the number of interventions independent variable;

[0133] The processing unit 802 is specifically used to: determine a target matrix equation based on the observation data of each observation object and the target regression equation, the target matrix equation including an independent variable matrix and a dependent variable matrix; determine at least one regression parameter in the target matrix equation based on the independent variable matrix and the dependent variable matrix; and determine the intervention effect parameters of each intervention event based on the at least one regression parameter.

[0134] In one embodiment, the processing unit 802 is specifically used to: determine the parameter values ​​corresponding to each variable in the target regression equation based on the observation data of each observation object, so as to obtain the auxiliary regression equation corresponding to each observation object; and determine the target matrix equation using the auxiliary regression equation corresponding to each observation object; wherein the independent variable matrix is ​​determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation corresponding to each observation object, and the dependent variable matrix is ​​determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation corresponding to each observation object.

[0135] In one embodiment, the processing unit 802 is specifically configured to: determine multiple observation times and multiple observation objects of the target application; for each of the multiple observation objects, acquire intervention data, response result data, and dynamic feature data corresponding to each observation object at each of the multiple observation times, wherein the intervention data corresponding to each observation time includes N intervention events corresponding to each observation time and the number of interventions for each intervention event; and use the intervention data, response result data, and dynamic feature data corresponding to each observation object at each observation time as the observation data for each observation object.

[0136] In one embodiment, the target regression equation includes a first regression equation, and the processing unit 802 is specifically configured to: determine the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the intervention frequency independent variable in the first regression equation based on N intervention events corresponding to each of the plurality of observation times and the intervention data corresponding to the target observation time for each observation object; label the target observation time according to the intervention start time of each intervention event included in the observation data and the plurality of observation times to obtain the parameter values ​​corresponding to the intervention time independent variable and the time effect independent variable in the first regression equation; label the dynamic feature data of each observation object corresponding to the target observation time to obtain the parameter values ​​corresponding to the dynamic effect independent variable in the first regression equation; and use the response result data of each observation object corresponding to the target observation time as the parameter values ​​corresponding to the dependent variable in the first regression equation.

[0137] In one embodiment, the processing unit 802 is specifically configured to: determine M observation objects corresponding to each observation time in multiple observation times for the target application, where M is an integer greater than 0; for each of the M observation objects, acquire static feature data of each observation object; acquire intervention data, response result data, and dynamic feature data of each observation object at the corresponding observation time, wherein the intervention data corresponding to the corresponding observation time includes N intervention events and the number of interventions for each intervention event; and use the static feature data of each observation object, as well as the intervention data, response result data, and dynamic feature data of each observation object at the corresponding observation time, as the observation data of each observation object.

[0138] In one embodiment, the target regression equation includes a second regression equation, and the processing unit 802 is specifically used to: determine the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the intervention number independent variable in the second regression equation based on N intervention events corresponding to each observation object at each corresponding observation time and the intervention data corresponding to each observation object at the target observation time; label the target observation time according to the intervention start time of each intervention event included in the observation data and the multiple observation times to obtain the parameter values ​​corresponding to the intervention time independent variable and the time effect independent variable in the second regression equation; label the static feature data of each observation object and the dynamic feature data of each observation object corresponding to the target observation time to obtain the parameter values ​​corresponding to the static effect independent variable and the dynamic effect independent variable in the second regression equation; and use the response result data of each observation object corresponding to the target observation time as the parameter value corresponding to the dependent variable in the second regression equation.

[0139] In one embodiment, the intervention effect parameters include a first regression parameter and a second regression parameter, and the processing unit 802 is specifically used to: determine the effectiveness evaluation parameters of each intervention event based on the first regression parameter and the second regression parameter included in the intervention effect parameters of each intervention event; and determine the target intervention event from the various intervention events based on the effectiveness evaluation parameters.

[0140] In one embodiment, the processing unit 802 is specifically configured to: determine the intervention effect correlation between the target intervention event and the number of interventions based on the first regression parameter included in the intervention effect parameter of the target intervention event; determine the target number of interventions based on the intervention effect correlation; and execute the target intervention event on the object using the target application according to the target number of interventions.

[0141] It is understood that the functions of each functional unit of the data processing device described in the embodiments of this application can be specifically implemented according to the methods in the above method embodiments, and the specific implementation process can be referred to the relevant descriptions in the above method embodiments, which will not be repeated here.

[0142] In this embodiment, observation data for each of the multiple observation objects of the target application can be obtained. The observation data includes N intervention events corresponding to each observation time within at least one observation time, and the number of interventions for each intervention event, where N is an integer greater than or equal to 0. Based on the observation data of each observation object and the target regression method, intervention effect parameters for each intervention event included in the observation data are determined. Target intervention events are determined based on the intervention effect parameters of each intervention event, and the target intervention events are executed on the objects using the target application. By employing this embodiment, the granularity of intervention effect measurement can be refined, and the intervention effects under different intervention events can be evaluated, thereby enabling the development of more effective intervention strategies for the objects using the target application.

[0143] like Figure 9 As shown, Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. The internal structure of the computer device 90 is as follows: Figure 9 As shown, it includes: one or more processors 901, a memory 902, and a communication interface 903. The processors 901, memory 902, and communication interface 903 can be connected via a bus 904 or other means; this embodiment of the application takes connection via bus 904 as an example.

[0144] The processor 901 (or CPU, Central Processing Unit) is the computing and control core of the computer device 90. It can parse various instructions within the computer device 90 and process various data. For example, the CPU can parse power-on / off commands sent by the user to the computer device 90 and control the computer device 90 to perform power-on / off operations; it can also transmit various interactive data between internal structures of the computer device 90, and so on. The communication interface 903 may optionally include standard wired interfaces or wireless interfaces (such as Wi-Fi, mobile communication interfaces, etc.), and is controlled by the processor 901 for sending and receiving data. The memory 902 is a storage device in the computer device 90 used to store computer programs and data. It is understood that the memory 902 here can include the computer device 90's built-in memory, or it can include extended memory supported by the computer device 90. The memory 902 provides storage space for the computer device 90's operating system, which may include, but is not limited to, Windows, Linux, Android, iOS, etc., and this application does not limit this. The processor 901 performs the following operations by running the computer program stored in the memory 902:

[0145] Obtain observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0;

[0146] Based on the observation data of each observation object and the target regression method, determine the intervention effect parameters of each intervention event included in the observation data;

[0147] The target intervention event is determined based on the intervention effect parameters of each intervention event, and the target intervention event is executed on the object using the target application.

[0148] In one embodiment, the processor 901 is specifically configured to: obtain a target regression equation, the target regression equation being determined based on the independent variable of the intervention event and the independent variable of the number of interventions; determine a target matrix equation based on the observation data of each observed object and the target regression equation, the target matrix equation including an independent variable matrix and a dependent variable matrix; determine at least one regression parameter in the target matrix equation based on the independent variable matrix and the dependent variable matrix, and determine the intervention effect parameter of each intervention event based on the at least one regression parameter.

[0149] In one embodiment, the processor 901 is specifically configured to: determine the parameter values ​​corresponding to each variable in the target regression equation based on the observation data of each observation object, so as to obtain the auxiliary regression equation corresponding to each observation object; and determine the target matrix equation using the auxiliary regression equation corresponding to each observation object; wherein the independent variable matrix is ​​determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation corresponding to each observation object, and the dependent variable matrix is ​​determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation corresponding to each observation object.

[0150] In one embodiment, the processor 901 is specifically configured to: determine multiple observation times and multiple observation objects of the target application; for each of the multiple observation objects, acquire intervention data, response result data, and dynamic feature data corresponding to each observation object at each of the multiple observation times, wherein the intervention data corresponding to each observation time includes N intervention events corresponding to each observation time and the number of interventions for each intervention event; and use the intervention data, response result data, and dynamic feature data corresponding to each observation object at each observation time as the observation data for each observation object.

[0151] In one embodiment, the target regression equation includes a first regression equation, and the processor 901 is specifically configured to: determine the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the intervention frequency independent variable in the first regression equation based on N intervention events corresponding to each of the plurality of observation times for each observed object and the intervention data corresponding to the target observation time for each observed object; label the target observation time according to the intervention start time of each intervention event included in the observation data and the plurality of observation times to obtain the parameter values ​​corresponding to the intervention time independent variable and the time effect independent variable in the first regression equation; label the dynamic feature data of each observed object corresponding to the target observation time to obtain the parameter values ​​corresponding to the dynamic effect independent variable in the first regression equation; and use the response result data of each observed object corresponding to the target observation time as the parameter values ​​corresponding to the dependent variable in the first regression equation.

[0152] In one embodiment, the processor 901 is specifically configured to: determine M observation objects corresponding to each observation time in multiple observation times for the target application, where M is an integer greater than 0; for each of the M observation objects, acquire static feature data of each observation object; acquire intervention data, response result data, and dynamic feature data of each observation object at the corresponding observation time, wherein the intervention data corresponding to the corresponding observation time includes N intervention events and the number of interventions for each intervention event; and use the static feature data of each observation object, as well as the intervention data, response result data, and dynamic feature data of each observation object at the corresponding observation time, as the observation data of each observation object.

[0153] In one embodiment, the target regression equation includes a second regression equation, and the processor 901 is specifically configured to: determine the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the intervention frequency independent variable in the second regression equation based on N intervention events corresponding to each corresponding observation time for each observed object and the intervention data corresponding to each observed object at the target observation time; label the target observation time according to the intervention start time of each intervention event included in the observation data and the multiple observation times to obtain the parameter values ​​corresponding to the intervention time independent variable and the time effect independent variable in the second regression equation; label the static feature data of each observed object and the dynamic feature data of each observed object corresponding to the target observation time to obtain the parameter values ​​corresponding to the static effect independent variable and the dynamic effect independent variable in the second regression equation; and use the response result data of each observed object corresponding to the target observation time as the parameter value corresponding to the dependent variable in the second regression equation.

[0154] In one embodiment, the intervention effect parameters include a first regression parameter and a second regression parameter, and the processor 901 is specifically configured to: determine the effectiveness evaluation parameters of each intervention event based on the first regression parameter and the second regression parameter included in the intervention effect parameters of each intervention event; and determine the target intervention event from the various intervention events based on the effectiveness evaluation parameters.

[0155] In one embodiment, the processor 901 is specifically configured to: determine the intervention effect correlation between the target intervention event and the number of interventions based on a first regression parameter included in the intervention effect parameters of the target intervention event; determine the target number of interventions based on the intervention effect correlation; and execute the target intervention event on the object using the target application according to the target number of interventions.

[0156] In specific implementations, the processor 901, memory 902, and communication interface 903 described in the embodiments of this application can execute the implementation method described in the data processing method provided in the embodiments of this application, or the implementation method described in the data processing device provided in the embodiments of this application, which will not be repeated here.

[0157] In this embodiment, observation data for each of the multiple observation objects of the target application can be obtained. The observation data includes N intervention events corresponding to each observation time within at least one observation time, and the number of interventions for each intervention event, where N is an integer greater than or equal to 0. Based on the observation data of each observation object and the target regression method, intervention effect parameters for each intervention event included in the observation data are determined. Target intervention events are determined based on the intervention effect parameters of each intervention event, and the target intervention events are executed on the objects using the target application. By employing this embodiment, the granularity of intervention effect measurement can be refined, and the intervention effects under different intervention events can be evaluated, thereby enabling the development of more effective intervention strategies for the objects using the target application.

[0158] This application also provides a computer-readable storage medium storing a computer program that, when run on a computer device, causes the computer device to perform any of the data processing methods described above. Specific implementations are described above and will not be repeated here.

[0159] This application also provides a computer program product, which includes a computer program or computer instructions. When executed by a processor, the computer program or computer instructions implement the steps of the data processing method provided in this application. The specific implementation method can be found in the foregoing description and will not be repeated here.

[0160] This application also provides a computer program comprising 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 provided in this application. Specific implementation details are provided above and will not be repeated here.

[0161] 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.

[0162] 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 related hardware. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0163] The above-disclosed embodiments are only some of the embodiments of this application, and should not be construed as limiting the scope of this application. Therefore, any equivalent changes made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A data processing method, characterized in that, The method includes: Obtain observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0; Obtain the target regression equation, which is determined based on the independent variables of intervention event, number of interventions, and intervention time; Based on the observation data of each observation object, determine the parameter values ​​corresponding to each variable in the target regression equation to obtain the auxiliary regression equation corresponding to each observation object; The target matrix equation is determined using the auxiliary regression equation corresponding to each observation object; the target matrix equation includes an independent variable matrix and a dependent variable matrix; the independent variable matrix is ​​determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation corresponding to each observation object, and the dependent variable matrix is ​​determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation corresponding to each observation object. Based on the independent variable matrix and the dependent variable matrix, at least one regression parameter in the target matrix equation is determined, and based on the at least one regression parameter, intervention effect parameters for each intervention event included in the observed data are determined. The target intervention event is determined based on the intervention effect parameters of each intervention event, and the target intervention event is executed on the object using the target application.

2. The method according to claim 1, characterized in that, The acquisition of observation data for each of the multiple observation objects in the target application includes: Determine multiple observation times and multiple observation objects for the target application; For each of the multiple observation objects, the intervention data, response result data, and dynamic feature data corresponding to each observation time in the multiple observation times are obtained. The intervention data corresponding to each observation time includes N intervention events corresponding to each observation time and the number of interventions for each intervention event. The intervention data, response result data, and dynamic feature data corresponding to each observation object at each observation time are used as the observation data for each observation object.

3. The method according to claim 2, characterized in that, The target regression equation includes a first regression equation, and determining the parameter values ​​corresponding to each variable in the target regression equation based on the observation data of each observation object includes: Based on the N intervention events corresponding to each observation time of each observation object in the plurality of observation times and the intervention data corresponding to the target observation time of each observation object, the parameter values ​​corresponding to the independent variables of the intervention events and the parameter values ​​corresponding to the independent variables of the number of interventions in the first regression equation are determined, and the target observation time is one of the plurality of observation times; Based on the intervention start time of each intervention event included in the observation data and the multiple observation times, the target observation time is labeled to obtain the parameter values ​​corresponding to the intervention time independent variable and the parameter values ​​corresponding to the time effect independent variable in the first regression equation. The dynamic feature data of each observed object corresponding to the target observation time are labeled to obtain the parameter values ​​corresponding to the dynamic effect independent variables in the first regression equation. The response result data of each observed object at the target observation time is used as the parameter value of the dependent variable in the first regression equation.

4. The method according to claim 1, characterized in that, The acquisition of observation data for each of the multiple observation objects in the target application includes: The target application is identified as having M observation objects corresponding to each observation time across multiple observation times, where M is an integer greater than 0; For each of the M observed objects, obtain the static feature data of each observed object; The intervention data, response result data, and dynamic feature data of each observed object at the corresponding observation time are obtained. The intervention data corresponding to the corresponding observation time includes N intervention events and the number of interventions for each intervention event. The static feature data of each observed object, as well as the intervention data, response result data, and dynamic feature data of each observed object at the corresponding observation time, are used as the observation data of each observed object.

5. The method according to claim 4, characterized in that, The target regression equation includes a second regression equation, and determining the parameter values ​​corresponding to each variable in the target regression equation based on the observation data of each observation object includes: Based on the N intervention events corresponding to each observation object at each corresponding observation time and the intervention data corresponding to each observation object at the target observation time, determine the parameter values ​​corresponding to the intervention event independent variable and the parameter values ​​corresponding to the intervention number independent variable in the second regression equation. The target observation time is one of the multiple observation times. Based on the intervention start time of each intervention event included in the observation data and the multiple observation times, the target observation time is labeled to obtain the parameter values ​​corresponding to the intervention time independent variable and the parameter values ​​corresponding to the time effect independent variable in the second regression equation. The static feature data of each observed object and the dynamic feature data of each observed object corresponding to the target observation time are labeled to obtain the parameter values ​​corresponding to the static effect independent variable and the parameter values ​​corresponding to the dynamic effect independent variable in the second regression equation. The response result data of each observed object at the target observation time is used as the parameter value of the dependent variable in the second regression equation.

6. The method as described in claim 1, characterized in that, The target regression equation includes the coefficients of a first interaction term and a second interaction term. The first interaction term is determined based on the intervention time and intervention event variables, and the second interaction term is determined based on the intervention frequency, intervention time, and intervention event variables. The target matrix equation also includes a matrix to be solved, which contains the coefficients of the first and second interaction terms. The target matrix equation is defined as follows: the result of matrix multiplication of the matrix to be solved and the independent variable matrix is ​​equal to the dependent variable matrix; the intervention effect parameters include a first regression parameter and a second regression parameter; Determining at least one regression parameter in the target matrix equation based on the independent variable matrix and the dependent variable matrix includes: Based on the definition of the independent variable matrix, the dependent variable matrix, and the target matrix equation, matrix solving is performed on the matrix to be solved in the target matrix equation to obtain the solution value corresponding to the coefficient of the first interaction term and the solution value corresponding to the coefficient of the second interaction term in the matrix to be solved. The solution value corresponding to the coefficient of the first interaction term is determined as the second regression parameter, and the solution value corresponding to the coefficient of the first interaction term is determined as the first regression parameter.

7. The method according to claim 6, characterized in that, The step of determining the target intervention event based on the intervention effect parameters of each intervention event includes: Based on the intervention effect parameters of each intervention event, including the first regression parameter and the second regression parameter, the effectiveness evaluation parameters of each intervention event are determined. The target intervention event is determined from the various intervention events based on the effectiveness evaluation parameters.

8. The method according to claim 7, characterized in that, The execution of the target intervention event on the object using the target application includes: Based on the first regression parameter included in the intervention effect parameters of the target intervention event, the correlation between the intervention effect of the target intervention event and the number of interventions is determined; The target number of interventions is determined based on the correlation between the intervention effects. The target intervention event is performed on the objects using the target application according to the target number of interventions.

9. A data processing apparatus, characterized in that, The device includes: An acquisition unit is used to acquire observation data for each of the multiple observation objects in the target application; wherein, the observation data includes N intervention events corresponding to each observation time in at least one observation time and the number of interventions for each intervention event, where N is an integer greater than or equal to 0; The processing unit is used to obtain a target regression equation, which is determined based on the independent variables of the intervention event, the number of interventions, and the intervention time; determine the parameter values ​​corresponding to each variable in the target regression equation based on the observation data of each observation object to obtain an auxiliary regression equation for each observation object; determine a target matrix equation using the auxiliary regression equation for each observation object; the target matrix equation includes an independent variable matrix and a dependent variable matrix; the independent variable matrix is ​​determined based on the parameter values ​​corresponding to the independent variables in the auxiliary regression equation for each observation object, and the dependent variable matrix is ​​determined based on the parameter values ​​corresponding to the dependent variables in the auxiliary regression equation for each observation object; determine at least one regression parameter in the target matrix equation based on the independent variable matrix and the dependent variable matrix, and determine the intervention effect parameters of each intervention event based on the at least one regression parameter; The processing unit is further configured to determine a target intervention event based on the intervention effect parameters of each intervention event, and to execute the target intervention event on the object using the target application.

10. A computer device, characterized in that, The computer device includes a memory, a communication interface, and a processor, wherein the memory, the communication interface, and the processor are interconnected; the memory stores a computer program, and the processor calls the computer program stored in the memory to implement the data processing method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the data processing method as described in any one of claims 1-8.

12. A computer program product, characterized in that, The computer program product includes a computer program or computer instructions, which, when executed by a processor, implement the data processing method as described in any one of claims 1-8.

Images