Construction of a causal inference model based on gradient boosting tree and related device

Abstract

This invention provides a method for constructing a causal inference model based on gradient boosting trees and related apparatus, relating to the field of artificial intelligence technology. The method includes acquiring intervention information, individual characteristics, and individual effects of multiple objects, wherein the individual effects include a first individual effect before the object is intervened and a second individual effect after the intervention; constructing a dataset from the intervention information, individual characteristics, and individual effects of multiple objects, wherein the intervention information, individual characteristics, and the first individual effect are input labels, and the second individual effect is the output label; constructing a causal inference model based on gradient boosting trees; and training and validating the causal inference model using the dataset to obtain a trained causal inference model. This invention, by combining the learning ability of GBDT with the objective of the causal inference model, achieves accurate prediction at the individual level, which helps optimize personalized decision-making and intervention strategies, thereby bringing greater value to various fields.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to the construction of a causal inference model based on a gradient boosting tree and related apparatus. Background Technology

[0002] In the field of machine learning, the Uplift model (causal inference model) is a causal inference method and a special predictive model designed to analyze the impact of intervention behaviors on individual responses. The Uplift model can be applied to various predictive domains. For example, during advertising campaigns, data such as ad impressions, clicks, and spending often fluctuate. These fluctuations are usually caused by user actions within the management system. Before the data fluctuations, users may have performed various types of actions, and these actions may not all be the cause of the fluctuations. Therefore, it is difficult for users to determine which actions were the primary cause of the data fluctuations. Summary of the Invention

[0003] The purpose of this invention is to provide a construction method and related apparatus for a causal inference model based on gradient boosting trees, so as to improve the above-mentioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0004] Firstly, this application provides a method for constructing a causal inference model based on gradient boosting trees, including:

[0005] The intervention information, individual characteristics, and individual effects of multiple subjects are obtained, wherein the individual effects include the first individual effect before the intervention and the second individual effect after the intervention.

[0006] A dataset is constructed from intervention information, individual characteristics, and individual effects of multiple objects. The intervention information, individual characteristics, and the first individual effect of the objects are the input labels, and the second individual effect is the output label.

[0007] A causal inference model based on gradient boosting tree is constructed, and the causal inference model is trained and validated using the dataset to obtain the trained causal inference model.

[0008] Secondly, this application also provides a device for constructing a causal inference model based on gradient boosting trees, comprising:

[0009] Information acquisition module: used to acquire intervention information, individual characteristics and individual effects of multiple objects, wherein the individual effects include the first individual effect before the object is subjected to intervention and the second individual effect after the object is subjected to intervention;

[0010] Dataset construction module: used to construct a dataset from intervention information, individual characteristics and individual effects of multiple objects, where the intervention information, individual characteristics and the first individual effect of the object are the input labels, and the second individual effect is the output label;

[0011] Causal inference model building module: used to build a causal inference model based on gradient boosting tree, and to train and validate the causal inference model using the dataset to obtain the trained causal inference model.

[0012] Thirdly, this application also provides an apparatus for constructing a causal inference model based on a gradient boosting tree, comprising:

[0013] Memory, used to store computer programs;

[0014] A processor is used to implement the steps of the method for constructing the gradient boosting tree-based causal inference model when executing the computer program.

[0015] Fourthly, this application also provides a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for constructing a causal inference model based on a gradient boosting tree.

[0016] The beneficial effects of this invention are as follows:

[0017] This invention combines causal inference models and Gradient Boosting Tree (GBDT) methods. Leveraging the powerful gradient boosting tree approach, it addresses the problem of causal inference, providing modelers with a robust and flexible method for causal inference and offering an effective framework and tool for predicting individual causal effects. By combining the learning capabilities of GBDT with the objectives of causal inference models, it achieves accurate predictions at the individual level, contributing to the optimization of personalized decision-making and intervention strategies, thereby bringing greater value to various fields.

[0018] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the construction method of the causal inference model based on gradient boosting tree described in the embodiments of the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of the device for constructing the causal inference model based on gradient boosting tree as described in an embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of the device structure for constructing the causal inference model based on gradient boosting tree as described in an embodiment of the present invention.

[0023] Marked in the image:

[0024] 800. Device for constructing causal inference models based on gradient boosting trees; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] Some embodiments of this application are modeled based on a structured causal model and construct the decision-making process based on a user-understood model combined with decision data. Through intervention methods or counterfactual reasoning, the expected effects of a decision can be verified.

[0028] Example 1:

[0029] This embodiment provides a method for constructing a causal inference model based on gradient boosting trees.

[0030] See Figure 1The figure shows that this method includes:

[0031] S1. Obtain intervention information t, individual characteristics x, and individual effects y for multiple subjects, wherein the individual effects include the first individual effect y of the subject before the intervention. before and the second individual effect y after intervention after ;

[0032] The individual characteristics may include at least one of the following: basic attributes, industry attributes, operation frequency, planned budget, and effective material scale; the individual effects may include the number of impressions in the most recent day, the number of consumptions in the most recent day, the number of clicks in the most recent day, etc.; the intervention information may include the placement of advertisements and the implementation of preferential strategies, etc.

[0033] Specifically, step S1 includes:

[0034] S11. Extract parameter data and outcome data of the object before intervention from the database;

[0035] S12. Use feature selection algorithms to extract features from object parameter data to obtain the individual characteristics of the object;

[0036] The feature selection algorithm can be Pearson coefficient, relief algorithm, or maximum correlation minimum redundancy algorithm, etc.

[0037] S13. Identify and remove distorted data in the result data. Perform data cleaning and data repair on the removed result data. The purpose of data cleaning is to filter out data that does not meet the requirements, such as incomplete data, erroneous data, and duplicate data, in order to remove some unnecessary data and reduce memory usage.

[0038] S14. Perform format, granularity, and vectorization transformations on the cleaned data to obtain the first volume effect of the object;

[0039] It should be noted that the method for obtaining the second volume effect is the same as that for the first volume effect. First, it is necessary to extract the result data of the object after the intervention from the database. Then, after processing the obtained data by removing distorted data, cleaning, repairing, formatting, granularizing, and vectorizing, the second volume effect is obtained.

[0040] S16. Identify and remove distorted data from the results data after intervention, and perform data cleaning and data repair on the removed results data;

[0041] Based on the above embodiments, this method further includes:

[0042] S2. Construct a dataset from the intervention information, individual characteristics, and individual effects of multiple objects, where the intervention information, individual characteristics, and the first individual effect of the objects are the input labels (x, t, y). b The second volumetric feature effect is the output label (y). a );

[0043] Specifically, step S2 includes:

[0044] S21. Divide the dataset into a training set and a validation set in an 8:2 ratio;

[0045] In this embodiment, the multiple objects obtained include individuals with different reactions, that is, groups that have a positive effect on the intervention. Therefore, based on the reaction results, all objects are divided into positive reaction individuals and negative reaction individuals. Positive reaction individuals are used as positive samples and negative reaction individuals are used as negative samples. A dataset is constructed from all positive samples and negative samples.

[0046] S22. Calculate the difference between the first and second individual effects for each training sample to obtain the intervention effect:

[0047] Δy=y a -y b (1)

[0048] In the formula, Δy represents the intervention effect.

[0049] Based on the above embodiments, this method further includes:

[0050] S3. Construct a causal inference model based on gradient boosting tree, and use the dataset to train and validate the causal inference model to obtain the trained causal inference model;

[0051] Specifically, step S3 includes:

[0052] S31. The training set and validation set are sequentially input into the causal inference model. After learning the causal relationship between individual characteristics and the second individual effect, the causal inference model outputs the predicted individual effect.

[0053] Specifically, step S31 includes:

[0054] S311. Initialize the cumulative effect Q0(x) i , t i In this embodiment, Q0(x) i , t i The initial value of ) is 0;

[0055] S312. Obtain the M decision trees in the causal inference model, and calculate the first residual effect using the intervention effect and cumulative effect of the training samples;

[0056] The i-th training sample (x) i , t i y ib ) and the corresponding intervention effect Δy i In the input causal inference model, all decision trees are trained iteratively, and the prediction error of the previous decision tree is used as the basis for training the next decision tree.

[0057] The calculation method for the first residual effect is as follows:

[0058] r i1 =Δy i -Q0(x i , t i )=Δy i (2)

[0059] S313. Train the first decision tree T1 with the first residual effect as the objective, and obtain the output effect of the first decision tree:

[0060] Q1(x i , t i )=Q0(x i , t i )+T1(x i , t i )=T1(x i , t i (3)

[0061] S314. Calculate the second residual effect r using the difference between the output effect of the first decision tree and the effect of the training samples. i2 :

[0062] r i2 =Δy i -Q1(x i , t i (4)

[0063] S315. Train the second decision tree with the second residual effect as the objective, and obtain the output effect Q2(x) of the second decision tree. i , t i );

[0064] Q2(x i , t i )=Q1(x i , t i )+T2(x i , t i (5)

[0065] S316. After all decision trees have been trained, the output effects of all decision trees are weighted and combined to obtain the predicted individual effects;

[0066] Specifically, determine whether the number of training iterations has reached M:

[0067] If so, stop training and calculate Q by weighting the output effects of all decision trees. M (x, t), that is, the prediction function f(x, t) = Q M (x, t).

[0068] S32. Calculate the loss value for predicting the individual effect and the second individual effect;

[0069] In this embodiment, the root mean square error (RMSE) is used to evaluate the loss between the predicted value and the true value. The calculation formula is as follows:

[0070]

[0071] In the formula: y′ i To predict individual effects, y ib This is the second body effect, and n is the number of samples in the validation set.

[0072] S33. Determine whether the loss value is greater than a preset loss standard:

[0073] If so, then the causal inference model training and completion are complete;

[0074] Otherwise, the causal inference model is iteratively optimized again to ensure that the loss value of the obtained causal inference model reaches the preset standard.

[0075] Example 2:

[0076] like Figure 2 As shown, this embodiment provides a device for constructing a causal inference model based on gradient boosting trees. The device includes:

[0077] Information acquisition module: used to acquire intervention information, individual characteristics and individual effects of multiple objects, wherein the individual effects include the first individual effect before the object is subjected to intervention and the second individual effect after the object is subjected to intervention;

[0078] Dataset construction module: used to construct a dataset from intervention information, individual characteristics and individual effects of multiple objects, where the intervention information, individual characteristics and the first individual effect of the object are the input labels, and the second individual effect is the output label;

[0079] Causal inference model building module: used to build a causal inference model based on gradient boosting tree, and to train and validate the causal inference model using the dataset to obtain the trained causal inference model.

[0080] Based on the above embodiments, the information acquisition module includes:

[0081] Data extraction unit: used to extract parameter data and result data of objects before intervention from the database;

[0082] Feature extraction unit: Used to extract features from object parameter data using feature selection algorithms to obtain the individual characteristics of the object.

[0083] Based on the above embodiments, the information acquisition module further includes:

[0084] Data cleaning unit: used to identify and remove distorted data in the result data, and to perform data cleaning and data repair on the removed result data;

[0085] Data preprocessing unit: Used to perform format, granularity, and vectorization transformations on the cleaned data to obtain the first volume effect of the object.

[0086] Based on the above embodiments, the dataset construction module includes:

[0087] Dataset partitioning unit: used to divide the dataset into training and validation sets in an 8:2 ratio;

[0088] The first calculation unit is used to calculate the difference between the first individual effect and the second individual effect for each training sample to obtain the intervention effect.

[0089] Based on the above embodiments, the dataset construction module further includes:

[0090] Individual effect prediction unit: used to input the training set and validation set into the causal inference model in sequence. After learning the causal relationship between individual characteristics and the second individual effect, the causal inference model outputs the predicted individual effect.

[0091] The second calculation unit is used to calculate the loss value of the predicted individual effect and the second individual effect;

[0092] First judgment unit: used to determine whether the loss value is greater than a preset loss standard.

[0093] If so, then the causal inference model training and completion are complete;

[0094] Otherwise, the causal inference model is iteratively optimized again to ensure that the loss value of the obtained causal inference model reaches the preset standard.

[0095] Based on the above embodiments, the individual effect prediction unit includes:

[0096] Initialization unit: used to initialize the cumulative effect;

[0097] The second calculation unit is used to obtain several decision trees in the causal inference model and calculate the first residual effect using the intervention effect and cumulative effect of the training samples.

[0098] The first training unit is used to train the first decision tree with the first residual effect as the target, and obtain the output effect of the first decision tree.

[0099] The third calculation unit is used to calculate the second residual effect using the difference between the output effect of the first decision tree and the effect of the training samples.

[0100] The second training unit is used to train the second decision tree with the second residual effect as the target, and obtain the output effect of the second decision tree.

[0101] The fourth calculation unit is used to calculate the predicted individual effect by weighting and combining the output effects of all decision trees after all decision trees have been trained.

[0102] It should be noted that the specific manner in which each module performs its operation in the apparatus described in the above embodiments has been described in detail in the embodiments of the method, and will not be elaborated here.

[0103] Example 3:

[0104] Corresponding to the above method embodiments, this embodiment also provides a device for constructing a causal inference model based on gradient boosting trees. The device for constructing a causal inference model based on gradient boosting trees described below and the method for constructing a causal inference model based on gradient boosting trees described above can be referred to in correspondence.

[0105] Figure 3 This is a block diagram illustrating a device 800 for constructing a causal inference model based on a gradient boosting tree, according to an exemplary embodiment. Figure 3 As shown, the device 800 for constructing the causal inference model based on gradient boosting trees may include a processor 801 and a memory 802. The device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0106] The processor 801 controls the overall operation of the gradient boosting tree-based causal inference model construction device 800 to complete all or part of the steps in the aforementioned gradient boosting tree-based causal inference model construction method. The memory 802 stores various types of data to support the operation of the gradient boosting tree-based causal inference model construction device 800. This data may include, for example, instructions for any application or method operating on the gradient boosting tree-based causal inference model construction device 800, as well as application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as a keyboard, mouse, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the gradient boosting tree-based causal inference model construction device 800 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0107] In an exemplary embodiment, the gradient boosting tree-based causal inference model construction device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the above-described gradient boosting tree-based causal inference model construction method.

[0108] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, these program instructions implement the steps of the method for constructing a gradient boosting tree-based causal inference model described above. For example, the computer-readable storage medium may be the memory 802 including the program instructions described above. These program instructions may be executed by the processor 801 of the gradient boosting tree-based causal inference model construction device 800 to complete the method for constructing the gradient boosting tree-based causal inference model described above.

[0109] Example 4:

[0110] Corresponding to the above method embodiments, this embodiment also provides a readable storage medium. The readable storage medium described below and the method for constructing a causal inference model based on gradient boosting trees described above can be referred to in relation to each other.

[0111] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for constructing a causal inference model based on a gradient boosting tree as described in the above method embodiments.

[0112] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.

[0113] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0114] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for constructing a causal inference model based on gradient boosting trees, characterized in that, include: The system acquires intervention information, individual characteristics, and individual effects for multiple objects. The individual effects include a first individual effect before the intervention and a second individual effect after the intervention. The individual characteristics include at least one of basic attributes, industry attributes, operation frequency, planned budget, and effective material scale. The individual effects include the number of impressions, the number of purchases, and the number of clicks in the most recent day. The intervention information includes the placement of advertisements and the implementation of promotional strategies. A dataset is constructed from intervention information, individual characteristics, and individual effects of multiple objects, where the intervention information, individual characteristics, and the first individual effect of the object are the input labels, and the second individual effect is the output label; A causal inference model based on gradient boosting tree is constructed, and the causal inference model is trained and validated using the dataset to obtain the trained causal inference model. The first volume effect of the object is obtained, including: Identify and remove distorted data from the result data, and perform data cleaning and data repair on the removed result data. The cleaned data is transformed in terms of format, granularity, and vectorization to obtain the first volume effect of the object; After constructing the dataset from intervention information, individual characteristics, and individual effects of multiple objects, it includes: The dataset was divided into a training set and a validation set in an 8:2 ratio. The intervention effect is obtained by calculating the difference between the first individual effect and the second individual effect for each training sample. The causal inference model was trained and validated using the dataset, including: The training set and validation set are sequentially input into the causal inference model. After learning the causal relationship between individual characteristics and the second individual effect, the causal inference model outputs the predicted individual effect. Calculate the loss value for predicting the individual effect and the second individual effect; Determine whether the loss value is greater than a preset loss standard: If so, then the causal inference model training and completion are complete; Otherwise, the causal inference model is iteratively optimized again to ensure that the loss value of the obtained causal inference model reaches the preset standard.

2. The method for constructing a causal inference model based on gradient boosting trees according to claim 1, characterized in that... To obtain the individual characteristics of an object, including: Extract parameter data and outcome data of the object from the database before intervention; Feature selection algorithms are used to extract features from object parameter data to obtain the individual characteristics of the object.

3. The method for constructing a causal inference model based on gradient boosting trees according to claim 1, characterized in that... After learning the causal relationship between individual characteristics and the second individual effect, the causal inference model outputs a predicted individual effect, including: Initialize cumulative effect; Obtain several decision trees from the causal inference model, and calculate the first residual effect using the intervention effect and cumulative effect of the training samples; The first decision tree is trained with the first residual effect as the target, and the output effect of the first decision tree is obtained; The second residual effect is calculated using the difference between the output effect of the first decision tree and the effect of the training samples; The second decision tree is trained with the second residual effect as the target, and the output effect of the second decision tree is obtained. Once all decision trees have been trained, the output effects of all decision trees are weighted and combined to obtain the predicted individual effects.

4. A device for constructing a causal inference model based on gradient boosting trees, characterized in that, include: Information Acquisition Module: Used to acquire intervention information, individual characteristics, and individual effects of multiple objects. The individual effects include a first individual effect before the intervention and a second individual effect after the intervention. The individual characteristics include at least one of basic attributes, industry attributes, operation frequency, planned budget, and effective material scale. The individual effects include the number of impressions, the number of purchases, and the number of clicks in the most recent day. The intervention information includes the placement of advertisements and the implementation of promotional strategies. Dataset construction module: used to construct a dataset from intervention information, individual characteristics and individual effects of multiple objects, where the intervention information, individual characteristics and the first individual effect of the object are the input labels and the second individual effect is the output label; Causal inference model construction module: used to construct a causal inference model based on gradient boosting tree, and to train and validate the causal inference model using the dataset to obtain the trained causal inference model; The information acquisition module also includes: Data cleaning unit: used to identify and remove distorted data in the result data, and to perform data cleaning and data repair on the removed result data; Data preprocessing unit: used to perform format, granularity, and vectorization transformations on the cleaned data to obtain the first volume effect of the object; The dataset building module includes: Dataset partitioning unit: used to divide the dataset into training and validation sets in an 8:2 ratio; The first calculation unit is used to calculate the difference between the first individual effect and the second individual effect for each training sample to obtain the intervention effect. The dataset building module also includes: Individual effect prediction unit: used to input the training set and validation set into the causal inference model in sequence. After learning the causal relationship between individual characteristics and the second individual effect, the causal inference model outputs the predicted individual effect. The second calculation unit is used to calculate the loss value of the predicted individual effect and the second individual effect; First judgment unit: used to determine whether the loss value is greater than a preset loss standard. If so, then the causal inference model training and completion are complete; Otherwise, the causal inference model is iteratively optimized again to ensure that the loss value of the obtained causal inference model reaches the preset standard.

5. The apparatus for constructing a causal inference model based on gradient boosting trees according to claim 4, characterized in that... The information acquisition module includes: Data extraction unit: used to extract parameter data and result data of objects before intervention from the database; Feature extraction unit: Used to extract features from object parameter data using feature selection algorithms to obtain the individual characteristics of the object.

6. The apparatus for constructing a causal inference model based on gradient boosting trees according to claim 4, characterized in that... The individual effect prediction unit includes: Initialization unit: used to initialize the cumulative effect; The second calculation unit is used to obtain several decision trees in the causal inference model and calculate the first residual effect using the intervention effect and cumulative effect of the training samples. The first training unit is used to train the first decision tree with the first residual effect as the target, and obtain the output effect of the first decision tree. The third calculation unit is used to calculate the second residual effect using the difference between the output effect of the first decision tree and the effect of the training samples. The second training unit is used to train the second decision tree with the second residual effect as the target, and obtain the output effect of the second decision tree. The fourth calculation unit is used to calculate the predicted individual effect by weighting and combining the output effects of all decision trees after all decision trees have been trained.

7. A device for constructing a causal inference model based on gradient boosting trees, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the method for constructing a causal inference model based on gradient boosting trees as described in any one of claims 1 to 3 when executing the computer program.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method for constructing a causal inference model based on a gradient boosting tree as described in any one of claims 1 to 3.

Images