Yield rate prediction method and device, equipment, storage medium and computer program product

Abstract

The invention relates to the technical field of data prediction, and discloses a yield prediction method, device and equipment, a storage medium and a computer program product, and the method comprises the steps: carrying out a random control test based on historical investment data, and obtaining random control test data; training an initial yield prediction model based on the random control test data to obtain a target yield prediction model, the initial yield prediction model being constructed based on a potential causal framework; and performing prediction according to the feature data of the user to be invested through the target return rate prediction model to obtain a current predicted return rate. According to the invention, the efficiency, accuracy and stability of yield prediction can be effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of yield prediction, and in particular to a yield prediction method, apparatus, device, storage medium and computer program product. Background Art

[0002] In the prior art, the gain analysis model method can be used to predict the change in revenue and cost, which corresponds to the numerator and denominator of the ROI (Return on Investment) formula. This requires two predictions, the first to predict the change in the numerator, the second to predict the change in the denominator, and then divide the two. The disadvantage of this is that two prediction models need to be trained and maintained, which is not as convenient as one prediction model; and the prediction of each model will have errors, and the division of the two will further amplify the errors, and when the denominator is small, the prediction results will become more unstable.

[0003] Therefore, how to effectively improve the efficiency, accuracy and stability of yield forecasting is a problem that needs to be solved urgently.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a rate of return prediction method, device, equipment and storage medium, aiming to solve the technical problem of how to effectively improve the efficiency, accuracy and stability of rate of return prediction.

[0006] To achieve the above purpose, the present application proposes a rate of return prediction method, which includes:

[0007] Conduct randomized controlled trials based on historical investment data to obtain randomized controlled trial data;

[0008] Training an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework;

[0009] The target rate of return prediction model is used to make predictions based on the characteristic data of the user to be invested, and the current predicted rate of return is obtained.

[0010] In one embodiment, the randomized controlled trial is performed based on the historical investment data to obtain the randomized controlled trial data, including:

[0011] Obtain a preset number of sample data and sample variables;

[0012] Randomly dividing the preset number of sample data into treatment group samples and control group samples according to the sample variables;

[0013] Based on the historical investment data, a control experiment is conducted on the treatment group samples and the control group samples to obtain randomized controlled trial data.

[0014] In one embodiment, the performing of a control experiment on the treatment group samples and the control group samples based on the historical investment data to obtain randomized controlled trial data includes:

[0015] Generate first characteristic data of the treatment group samples and second characteristic data of the control group samples based on the historical investment data;

[0016] Generate first business data of the treatment group samples and second business data of the control group samples based on the first feature data and the second feature data respectively through a preset business model;

[0017] The randomized controlled trial data is determined according to the first business data and the second business data.

[0018] In one embodiment, the training of the initial rate of return prediction model based on the randomized controlled trial data to obtain the target rate of return prediction model includes:

[0019] Construct an initial yield prediction model based on a potential causal framework;

[0020] The initial rate of return prediction model is trained by using the randomized controlled trial data to obtain a predicted loss value;

[0021] A target rate of return prediction model is determined based on the predicted loss value.

[0022] In one embodiment, the training of the initial rate of return prediction model using the randomized controlled trial data to obtain a predicted loss value includes:

[0023] constructing a data set based on the randomized controlled trial data;

[0024] Inputting the characteristics of each sample in the data set into the initial yield prediction model to obtain a historical predicted yield;

[0025] Determine a reference rate of return based on the business data of each sample in the data set;

[0026] A predicted loss value is determined based on the historical predicted rate of return, the reference rate of return and a convex loss function.

[0027] In one embodiment, the training of the initial rate of return prediction model using the randomized controlled trial data to obtain a predicted loss value includes:

[0028] constructing a data set based on the randomized controlled trial data;

[0029] Inputting the characteristics of each sample in the data set into the initial yield prediction model to obtain a historical predicted yield;

[0030] Determine a reference rate of return based on the business data of each sample in the data set;

[0031] A predicted loss value is determined based on the historical predicted rate of return, the reference rate of return and a convex loss function.

[0032] In addition, to achieve the above purpose, the present application also proposes a yield prediction device, which includes:

[0033] An experiment module, used to conduct a randomized controlled experiment based on historical investment data to obtain randomized controlled experiment data;

[0034] A training module, used for training an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework;

[0035] The prediction module is used to predict the target yield rate according to the characteristic data of the user to be invested through the target yield rate prediction model to obtain the yield rate.

[0036] In addition, to achieve the above-mentioned purpose, the present application also proposes a yield prediction device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the yield prediction method described above.

[0037] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the yield prediction method described above are implemented.

[0038] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the yield prediction method described above.

[0039] The present application provides a rate of return prediction method. The present application first conducts a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; trains an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework; and uses the target rate of return prediction model to perform predictions based on characteristic data of users to be invested to obtain a current predicted rate of return, thereby effectively improving the efficiency, accuracy and stability of rate of return predictions.

[0040] In summary, the present application uses the randomized controlled trial data obtained from the randomized controlled trial to train the initial yield prediction model, thereby effectively improving the stability and prediction accuracy of the target yield prediction model, and then uses the target yield prediction model to quickly and accurately obtain the current predicted yield based on the characteristic data of the user to be invested, thereby overcoming the technical defects of the existing yield prediction methods in terms of poor efficiency and accuracy, and can effectively improve the efficiency, accuracy and stability of the yield prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0043] Figure 1 A flow chart of the first embodiment of the yield prediction method of the present application;

[0044] Figure 2 A flow chart of the second embodiment of the yield prediction method of this application;

[0045] Figure 3 This is a schematic diagram of the module structure of the yield prediction device of the embodiment of the present application;

[0046] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the yield prediction method of the embodiment of the present application.

[0047] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0048] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0049] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0050] The main solution of the embodiment of the present application is: conduct a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; train an initial yield prediction model based on the randomized controlled trial data to obtain a target yield prediction model, wherein the initial yield prediction model is constructed based on a potential causal framework; and use the target yield prediction model to make predictions based on the characteristic data of the users to be invested to obtain the current predicted yield.

[0051] In the prior art, the gain analysis model method can be used to predict the changes in revenue and cost, which correspond to the numerator and denominator of the ROI formula. This requires two predictions, the first to predict the change in the numerator, the second to predict the change in the denominator, and then divide the two. The disadvantage of this is that it requires training and maintaining two prediction models, which is not as convenient as one prediction model; and each model's prediction will have errors, and the division of the two will further amplify the errors, and when the denominator is small, the prediction results will become more unstable. Therefore, how to effectively improve the efficiency, accuracy and stability of yield prediction is a problem that needs to be solved urgently.

[0052] This application uses the randomized controlled trial data obtained from the randomized controlled trial to train the initial yield prediction model, thereby effectively improving the stability and prediction accuracy of the target yield prediction model, and then uses the target yield prediction model to quickly and accurately obtain the current predicted yield based on the characteristic data of the user to be invested, thereby overcoming the technical defects of the existing yield prediction methods in terms of poor efficiency and accuracy, and can effectively improve the efficiency, accuracy and stability of yield prediction.

[0053] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a yield prediction device, etc. The following takes the yield prediction device as an example to illustrate this embodiment and the following embodiments.

[0054] Based on this, the present application embodiment provides a rate of return prediction method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the yield prediction method of the present application.

[0055] In this embodiment, the rate of return prediction method includes steps S10 to S30:

[0056] Step S10, performing a randomized controlled trial based on historical investment data to obtain randomized controlled trial data.

[0057] It should be noted that the randomized controlled trial (RCT) is an important experimental design used for causal inference. It eliminates the influence of confounding variables by randomly assigning individuals to treatment and control groups to ensure that the estimation of treatment effects is unbiased.

[0058] It can be understood that historical investment data refers to the investment result data of different users under different investment behaviors, which may include user characteristics, investment behaviors, and investment results. Among them, investment behavior can be whether to accept certain marketing measures, and investment results can be the income and cost after marketing.

[0059] It is worth mentioning that randomized controlled trial data, namely RCT data, are data obtained through randomized controlled trials. These data can reflect the impact of different investment strategies on returns.

[0060] In a feasible implementation, in order to eliminate the influence of confounding variables and improve data accuracy, step S10 may include: obtaining a preset number of sample data and sample variables; randomly dividing the preset number of sample data into treatment group samples and control group samples according to the sample variables; conducting a control test on the treatment group samples and the control group samples based on the historical investment data to obtain randomized controlled test data.

[0061] It should be noted that the sample data can be determined according to business needs, that is, the number of samples to be generated is determined according to business needs. For example, for a user group of a specific product, 10,000 sample data may need to be generated. This embodiment does not impose specific restrictions on this.

[0062] It can be understood that sample variables are used to distinguish the characteristics or conditions of different samples, including user characteristics, processing variables and result variables, among which user characteristics can be the user's age, gender, region, historical consumption behavior, etc., processing variables are whether to accept certain marketing measures, and result variables can be the income and cost after marketing.

[0063] It is worth noting that according to the different treatment variables, it can be divided into a treatment group and a control group. The treatment group is the sample that receives marketing, and the control group is the sample that does not receive marketing. The generated samples are randomly assigned to the treatment group and the control group. For example, if there are 10,000 samples, 5,000 samples can be randomly assigned to the treatment group and the other 5,000 samples to the control group.

[0064] It is worth noting that in actual business scenarios, user behavior may be affected by many factors, such as the user's age, income level, historical purchase records, etc. Through random allocation processing, randomized controlled trial data can eliminate the influence of confounding variables and improve data accuracy.

[0065] In a feasible implementation, in order to accurately estimate the causal effect of an individual and improve data accuracy, a controlled experiment is conducted on the treatment group samples and the control group samples based on the historical investment data to obtain randomized controlled trial data, including: generating first characteristic data of the treatment group samples and second characteristic data of the control group samples based on the historical investment data; generating first business data of the treatment group samples and second business data of the control group samples based on the first characteristic data and the second characteristic data respectively through a preset business model; and determining the randomized controlled trial data based on the first business data and the second business data.

[0066] It should be noted that the characteristic data generated for each sample can be simulated based on user data in actual business, that is, characteristic data such as age, income, and past purchase records are generated for each user based on historical investment data.

[0067] It can be understood that the preset business model can be a model built based on actual business scenarios. For example, it can be a regression model, classification model, etc. built based on historical data. Through these models, the user's behavior and results under different investment strategies can be simulated to generate corresponding business data.

[0068] It is worth noting that business data can reflect the actual performance of users under different investment strategies, that is, the revenue and cost data after marketing.

[0069] In a specific implementation, the characteristic data of each sample is processed and allocated according to the business model to generate potential results for each sample. For example, the business model is used to simulate the post-marketing revenue and cost data based on user characteristics and processing conditions. For the treatment group samples, Y(1), i.e., the first business data, is generated, and for the control group samples, Y(0), i.e., the second business data, is generated.

[0070] It is worth noting that in randomized controlled trials, by comparing the results of the treatment group and the control group, the individual causal effect can be accurately estimated, providing an important basis for the accurate prediction of the rate of return.

[0071] Step S20, training an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework.

[0072] It should be noted that the rate of return, or return on investment (ROI), refers to the ratio between investment income and investment cost, and is usually used to measure the benefits of an investment. The rate of return prediction model refers to the one used to predict the rate of return on investment (ROI). The initial rate of return prediction model is constructed based on a potential causal framework, so that it can more accurately predict the rate of return under different investment strategies. The target rate of return prediction model is a model that meets the requirements for stability and prediction accuracy after training and optimizing the initial rate of return prediction model through randomized controlled trial data.

[0073] It can be understood that in causal inference, the potential causal framework (Potential Outcomes Fra mework) is a theoretical model used to estimate individual treatment effects. Here, "treatment" (Treatment) refers to the intervention measures imposed on an individual, for example, advertising or discounts to customers in a marketing scenario. The corresponding symbol is T, indicating whether the individual accepts the treatment, T = 1 means acceptance of the treatment, and T = 0 means non-acceptance of the treatment. Potential outcomes (Potential Outcomes) are the core concept in causal inference, which represent the possible results of an individual under different treatment conditions. For example, in the scenario of rate of return estimation, the potential outcome Y (1) represents the income or cost when the treatment (such as marketing) is accepted, while Y (0) represents the income or cost when the treatment is not accepted. In the potential causal framework, the causal effect (Causal Effect) of an individual is as follows:

[0074] Causal Effect = Y(1)-Y(0)

[0075] Among them, Y(1) represents the income or cost when receiving treatment (such as marketing), and Y(0) represents the income or cost when not receiving treatment.

[0076] It is worth noting that [feature, result, processing] Let X∈R d , X represents the eigenvector, which is realized as x i Although marketing activities can bring incremental revenue, they also generate significant costs. r ,Y c Represent the income result and cost result respectively, and their realization is Let T∈0,1 represent the processing, and its implementation is t_i. Let (Y r (1),Y r (0)) and (Y c (1),Y c (0)) represent the potential results when the individual accepts or does not accept the treatment. Define the change in income τ r (x i ) and cost change τ c (xi ) is the conditional average treatment effect, as follows:

[0077] τ * (x i )=E[Y * (1)-Y * (0)|X=x i ],*∈{r,c}.

[0078] Among them, τ * (x i ) refers to the change in revenue or cost, Y * (1) is the income result or cost result when receiving the treatment, Y * (0) is the income result or cost result when no processing is accepted.

[0079] Return on investment ROI i It refers to the change in revenue from marketing to an individual divided by the change in cost, as shown in the following formula:

[0080]

[0081] Among them, roi i is the return on investment, τ r (x i ) is the change in return, τ c (x i ) is the cost change. In actual business, the cost change is generally positive (that is, the cost is generally increasing), so τ c (x i )>0.

[0082] Step S30, predicting based on the characteristic data of the user to be invested by the target rate of return prediction model to obtain the current predicted rate of return.

[0083] It should be noted that the characteristic data of the user to be invested is input into the target rate of return prediction model, and the output of the model is the current predicted rate of return. In this embodiment, the ROI can be predicted by only one neural network model, without the need for real labeled data of the ROI, and without the need to divide the predicted values ​​of the two models to obtain the predicted value of the ROI, which effectively improves the efficiency and accuracy of the rate of return prediction.

[0084] It is understandable that the prediction results of the target rate of return prediction model can provide important decision support for investors. By analyzing the predicted rate of return, investors can evaluate the potential returns of different investment strategies and make more informed investment choices. In addition, the model can also help investors identify potential risks and avoid unnecessary losses.

[0085] The present embodiment provides a rate of return prediction method. The present embodiment first conducts a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; trains an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework; and uses the target rate of return prediction model to perform predictions based on characteristic data of users to be invested to obtain a current predicted rate of return, thereby effectively improving the efficiency, accuracy and stability of rate of return predictions.

[0086] In summary, this embodiment uses the randomized controlled trial data obtained from the randomized controlled trial to train the initial yield prediction model, thereby effectively improving the stability and prediction accuracy of the target yield prediction model, and then uses the target yield prediction model to quickly and accurately obtain the current predicted yield based on the characteristic data of the user to be invested, thereby overcoming the technical defects of the existing yield prediction methods in terms of poor efficiency and accuracy, and can effectively improve the efficiency, accuracy and stability of the yield prediction.

[0087] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can refer to the above introduction, and will not be repeated later. Figure 2 , the step S20 further includes steps S201-S203:

[0088] Step S201, constructing an initial yield prediction model based on a potential causal framework.

[0089] It should be noted that the initial yield prediction model is a neural network model built based on a potential causal framework. Users can make independent decisions on the hyperparameters of the initial yield prediction model, such as the number of neural network layers, number of neurons, and learning rate, based on actual business.

[0090] Step S202: training the initial rate of return prediction model using the randomized controlled trial data to obtain a predicted loss value.

[0091] It should be noted that the randomized controlled trial data is used to train the initial rate of return prediction model, and the predicted value of ROI is directly optimized through a customized convex loss function.

[0092] It is understandable that the initial yield prediction model will learn how to predict individual ROIs based on user characteristics, treatment situations, and observed results, thereby achieving accurate and stable ROI predictions.

[0093] In a feasible implementation, to improve the model stability and generalization ability, step S202 may include: constructing a data set based on the randomized controlled trial data; inputting the characteristics of each sample in the data set into the initial yield prediction model to obtain a historical predicted yield; determining a reference yield based on the business data of each sample in the data set; and determining a predicted loss value based on the historical predicted yield, the reference yield, and a convex loss function.

[0094] It should be noted that the convex loss function is a type of loss function in machine learning and statistical learning theory. Within the domain of a convex function, any local minimum is also a global minimum. This property makes it easier to solve the optimization problem and avoids falling into the problem of local optimal solutions.

[0095] Understandably, the dataset The size of is N, where the number of samples that are processed or not is N1 and N0 respectively. Optimizing the initial yield prediction model by using the convex loss function can effectively improve the stability and prediction accuracy of the yield prediction model. The convex loss function is as follows:

[0096]

[0097] Among them, q i is an unbiased estimator of the ROI prediction value, h(x i ) is a rate of return prediction model with user-determined hyperparameters. The parameters of the neural network in the model are obtained by model training. Its input x i is the characteristic of the i-th sample, σ(s i ) is the sigmoid function (activation function), which ensures that its output value q i The value range is (0,1), L(s) is the convex loss function, and ln is the logarithm with base e.

[0098] Step S203, determining a target rate of return prediction model according to the predicted loss value.

[0099] It should be noted that when the loss function of the rate of return prediction model is a convex function, when the convex loss function converges through model training, the output of the predicted loss value is the unbiased estimator of the ROI prediction value.

[0100] It is understandable that causal inference can be used to simplify the loss function as follows:

[0101]

[0102] Among them, N1 is the number of samples to be processed, q i is an unbiased estimator of the ROI prediction value, Yi r Receive processed income results, is the realization of income results and T is the processing.

[0103] E[Y i r (1)lnκ(q i )|T i =1] and E[Y i r (1)lnκ(q i )] are the consistency assumption and the exchangeability assumption respectively. The consistency assumption in causal inference means that if an individual receives a certain treatment, that is, T in the equation i =1, then the observed outcome Y i r It should be compared with the potential outcome Y if the individual receives the treatment. i r (1) Consistency. This means “If I accept this treatment, my outcome will be Y.” i r Should and assuming I accept this treatment, the potential result Y i r (1) The same.

[0104] The exchangeability assumption means that if different treatment groups are exchanged, the results should remain the same. Exchangeability ensures that the different groups are identical in all relevant aspects before treatment, except for the treatment itself. Therefore, removing the denominator of the equation will not change the value of its numerator, as shown below:

[0105]

[0106] Among them, X represents the feature vector, τ r is the change in revenue, N is the size of the data set, q i is an unbiased estimator of the ROI prediction value, Y i r Receive processed income results, is the realization of income results, T is the treatment, lnκ(q i ) can be replaced by Or ln(1-q i )

[0107] Therefore, the loss function can be simplified as:

[0108]

[0109] The loss function is a convex function for the following reasons:

[0110]

[0111] Therefore, the second-order derivative is as follows:

[0112]

[0113] Among them, q i =σ(s i ), q i ∈(0,1) and τ c (x i )>0, so we can get and That is, the second-order derivative matrix is ​​a Hessian matrix, so the loss function L(s) is a convex function.

[0114] In a feasible implementation, in order to improve the stability of the yield prediction model, step S203 may include: determining whether the trained initial yield prediction model has converged according to the predicted loss value; when the trained initial yield prediction model has converged, determining the target yield prediction model according to the trained initial yield prediction model; when the trained initial yield prediction model has not converged, optimizing the trained initial yield prediction model according to the predicted loss value until the trained initial yield prediction model has converged.

[0115] It should be noted that when training the yield prediction model, because the loss function is a convex function, the loss function has a minimum value, and there is only one minimum value, that is, the global minimum value. Then, when the loss function L(s) converges, that is, when the loss function reaches the global minimum value, its first-order derivative is 0, that is, Further we can get That is, the neural network The sigmoid output value q i It is an unbiased estimator of ROI.

[0116] It can be understood that convergence means that the update of model parameters gradually decreases during the training process until a stable state is reached, at which time the model's prediction loss value for the training data no longer decreases significantly or the decrease is lower than a preset threshold.

[0117] In the specific implementation, a maximum number of iterations is set. When the number of iterations reaches this value, the training process will be stopped even if the model has not fully converged to avoid overfitting and waste of computing resources.

[0118] In order to further improve the generalization ability of the model, in addition to checking whether the model converges, the cross-validation method can also be used to evaluate the performance of the model.

[0119] In this embodiment, the initial yield prediction model is trained by randomized controlled trial data, and then the loss value is predicted to optimize the initial yield prediction model to obtain a target yield prediction model, which can effectively improve the stability and prediction accuracy of the yield prediction model.

[0120] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the yield prediction method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0121] This application also provides a yield prediction device, please refer to Figure 3 , the yield prediction device comprises:

[0122] The test module 10 is used to conduct a randomized controlled trial based on historical investment data to obtain randomized controlled trial data.

[0123] The training module 20 is used to train an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework.

[0124] The prediction module 30 is used to predict the rate of return based on the characteristic data of the user to be invested through the target rate of return prediction model to obtain the rate of return.

[0125] The present embodiment provides a yield prediction device. The present embodiment conducts a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; trains an initial yield prediction model based on the randomized controlled trial data to obtain a target yield prediction model, wherein the initial yield prediction model is constructed based on a potential causal framework; and uses the target yield prediction model to perform predictions based on characteristic data of users to be invested to obtain a current predicted yield, which can effectively improve the efficiency, accuracy and stability of yield predictions.

[0126] In summary, this embodiment uses the randomized controlled trial data obtained from the randomized controlled trial to train the initial yield prediction model, thereby effectively improving the stability and prediction accuracy of the target yield prediction model, and then uses the target yield prediction model to quickly and accurately obtain the current predicted yield based on the characteristic data of the user to be invested, thereby overcoming the technical defects of the existing yield prediction methods in terms of poor efficiency and accuracy, and can effectively improve the efficiency, accuracy and stability of the yield prediction.

[0127] Optionally, the experimental module 10 is also used to obtain a preset number of sample data and sample variables; randomly divide the preset number of sample data into treatment group samples and control group samples according to the sample variables; and conduct a control experiment on the treatment group samples and the control group samples based on the historical investment data to obtain randomized controlled experiment data.

[0128] Optionally, the experimental module 10 is also used to generate first characteristic data of the treatment group samples and second characteristic data of the control group samples based on the historical investment data; generate first business data of the treatment group samples and second business data of the control group samples based on the first characteristic data and the second characteristic data respectively through a preset business model; and determine randomized controlled trial data based on the first business data and the second business data.

[0129] Optionally, the training module 20 is also used to construct an initial yield prediction model based on a potential causal framework; train the initial yield prediction model through the randomized controlled trial data to obtain a predicted loss value; and determine a target yield prediction model based on the predicted loss value.

[0130] Optionally, the training module 20 is also used to construct a data set based on the randomized controlled trial data; input the characteristics of each sample in the data set into the initial yield prediction model to obtain a historical predicted yield; determine a reference yield based on the business data of each sample in the data set; and determine a predicted loss value based on the historical predicted yield, the reference yield and a convex loss function.

[0131] Optionally, the training module 20 is also used to determine whether the trained initial yield prediction model has converged according to the predicted loss value; when the trained initial yield prediction model has converged, determine the target yield prediction model according to the trained initial yield prediction model; when the trained initial yield prediction model has not converged, optimize the trained initial yield prediction model according to the predicted loss value until the trained initial yield prediction model has converged.

[0132] The yield prediction device provided by the present application adopts the yield prediction method in the above embodiment, which can solve the technical problem of how to effectively improve the efficiency, accuracy and stability of yield prediction. Compared with the prior art, the beneficial effects of the yield prediction device provided by the present application are the same as the beneficial effects of the yield prediction method provided by the above embodiment, and the other technical features in the yield prediction device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0133] The present application provides a yield prediction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the yield prediction method in the above-mentioned embodiment one.

[0134] Reference below Figure 4 , which shows a schematic diagram of the structure of a yield prediction device suitable for implementing the embodiment of the present application. The yield prediction device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The yield prediction device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0135] like Figure 4 As shown, the yield prediction device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the yield prediction device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the yield prediction device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a yield prediction device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0136] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0137] The yield prediction device provided by the present application adopts the yield prediction method in the above embodiment, which can solve the technical problem of how to effectively improve the efficiency, accuracy and stability of yield prediction. Compared with the prior art, the beneficial effects of the yield prediction device provided by the present application are the same as the beneficial effects of the yield prediction method provided by the above embodiment, and the other technical features in the yield prediction device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0138] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0139] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0140] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the yield prediction method in the above-mentioned embodiment.

[0141] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0142] The above-mentioned computer-readable storage medium may be included in the yield prediction device; or it may exist independently without being assembled into the yield prediction device.

[0143] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the yield prediction device, the yield prediction device: conducts a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; trains an initial yield prediction model based on the randomized controlled trial data to obtain a target yield prediction model, wherein the initial yield prediction model is constructed based on a potential causal framework; and uses the target yield prediction model to make predictions based on the characteristic data of the user to be invested to obtain the current predicted yield.

[0144] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0145] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0146] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0147] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned rate of return prediction method, and can solve the technical problem of how to effectively improve the efficiency, accuracy, and stability of rate of return prediction. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the rate of return prediction method provided in the above-mentioned embodiment, and will not be repeated here.

[0148] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned rate of return prediction method when executed by a processor.

[0149] The computer program product provided by this application can solve the technical problem of how to effectively improve the efficiency, accuracy and stability of rate of return prediction. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the rate of return prediction method provided by the above embodiment, which will not be repeated here.

[0150] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A rate of return prediction method, characterized in that: The method comprises: Conduct randomized controlled trials based on historical investment data to obtain randomized controlled trial data; Training an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework; The target rate of return prediction model is used to make predictions based on the characteristic data of the user to be invested, and the current predicted rate of return is obtained.

2. The method according to claim 1, characterized in that The randomized controlled trial is conducted based on the historical investment data to obtain randomized controlled trial data, including: Obtain a preset number of sample data and sample variables; Randomly dividing the preset number of sample data into treatment group samples and control group samples according to the sample variables; Based on the historical investment data, a control experiment is conducted on the treatment group samples and the control group samples to obtain randomized controlled trial data.

3. The method according to claim 2, characterized in that The method of performing a control experiment on the treatment group samples and the control group samples based on the historical investment data to obtain randomized controlled trial data includes: Generate first characteristic data of the treatment group samples and second characteristic data of the control group samples based on the historical investment data; Generate first business data of the treatment group samples and second business data of the control group samples based on the first feature data and the second feature data respectively through a preset business model; The randomized controlled trial data is determined according to the first business data and the second business data.

4. The method according to claim 1, characterized in that The step of training the initial rate of return prediction model based on the randomized controlled trial data to obtain the target rate of return prediction model includes: Construct an initial yield prediction model based on a potential causal framework; The initial rate of return prediction model is trained by using the randomized controlled trial data to obtain a predicted loss value; A target rate of return prediction model is determined based on the predicted loss value.

5. The method according to claim 4, characterized in that The initial rate of return prediction model is trained by the randomized controlled trial data to obtain a predicted loss value, including: constructing a data set based on the randomized controlled trial data; Inputting the characteristics of each sample in the data set into the initial yield prediction model to obtain a historical predicted yield; Determine a reference rate of return based on the business data of each sample in the data set; A predicted loss value is determined based on the historical predicted rate of return, the reference rate of return and a convex loss function.

6. The method according to claim 4, characterized in that Determining a target rate of return prediction model according to the predicted loss value includes: Determining whether the trained initial rate of return prediction model has converged according to the predicted loss value; When the trained initial rate of return prediction model has converged, determining a target rate of return prediction model according to the trained initial rate of return prediction model; When the trained initial rate of return prediction model has not converged, the trained initial rate of return prediction model is optimized according to the predicted loss value until the trained initial rate of return prediction model has converged.

7. A yield prediction device, characterized in that: The yield prediction device comprises: An experimental module, used to conduct a randomized controlled trial based on historical investment data to obtain randomized controlled trial data; A training module, used for training an initial rate of return prediction model based on the randomized controlled trial data to obtain a target rate of return prediction model, wherein the initial rate of return prediction model is constructed based on a potential causal framework; The prediction module is used to predict the target yield rate according to the characteristic data of the user to be invested through the target yield rate prediction model to obtain the yield rate.

8. A yield prediction device, characterized in that: The yield prediction device includes: a memory, a processor, and a yield prediction program stored in the memory and executable on the processor, wherein the yield prediction program is configured to implement the yield prediction method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium stores a yield prediction program, which, when executed by a processor, implements the yield prediction method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The computer program product comprises a rate of return prediction program, which, when executed by a processor, implements the steps of the rate of return prediction method according to any one of claims 1 to 6.

Images