Customer asset promotion prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses a customer asset promotion prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: dividing the promotion amount of target customer assets into a plurality of asset intervals, and obtaining a corresponding prediction model based on the training of each asset interval; obtaining a prediction probability of a target customer in each asset interval according to each prediction model; and affiliating the target customer to the asset interval where the maximum prediction probability is located, and obtaining the final prediction probability and the asset promotion amount interval of the target customer. According to the invention, identification of potential asset promotion clients is realized.

Description

Technical Field

[0001] The present application relates to the field of asset prediction technology, and in particular to a method, device, electronic device, and storage medium for predicting customer asset enhancement. Background Art

[0002] In the financial and banking fields, accurately predicting the probability and amount of a customer's asset increase makes it easier for marketers to formulate effective marketing strategies, enhance customer experience, and increase corporate profits.

[0003] There are two main methods for predicting customer asset improvement:

[0004] First, it mainly relies on the experience and judgment of business personnel. The disadvantages of this method are: the subjectivity of business personnel's experience: different business personnel have different experiences, judgment standards and risk preferences, which may lead to subjectivity and inconsistency in the prediction results.

[0005] Second, use statistical models, such as linear regression, logistic regression, etc. These methods build prediction models by collecting basic customer information (such as age, gender, occupation, etc.), transaction data (such as transaction frequency, transaction amount, etc.) and asset information (such as deposit balance, loan status, etc.). The disadvantages of this method are: insufficient model prediction accuracy: because the rules or algorithms used are relatively simple, it is difficult to capture the complex relationships and nonlinear characteristics in the data, resulting in limited prediction accuracy and failure to accurately reflect the actual situation of customer asset changes. Summary of the invention

[0006] The embodiments of the present application provide a customer asset enhancement prediction method, device, electronic device, and storage medium to identify potential asset enhancement customers.

[0007] The present application embodiment adopts the following technical solutions:

[0008] In a first aspect, an embodiment of the present application provides a method for predicting customer asset improvement, wherein the method is performed by an asset management server, and the prediction method includes:

[0009] Divide the target customer's asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training;

[0010] Obtaining the predicted probability of the target customer in each asset range according to each of the prediction models;

[0011] Assign the target customer to the asset range with the maximum predicted probability, and obtain the target customer's final predicted probability and asset improvement amount range.

[0012] In some embodiments, before obtaining the corresponding prediction model based on each asset interval training, the method further includes:

[0013] Acquire information that affects changes in the target customer's assets during a preset observation period, where the information that affects changes in the customer's assets includes at least one of the following: basic information, attribute information, account information, asset information, transaction information, and preference information;

[0014] The increase amount of the target customer's assets is divided into k asset intervals and target variables 1, ..., k are set respectively. Meanwhile, the target variables of the target customer's assets that are not increased to the above asset intervals are set to 0.

[0015] In some embodiments, before obtaining the corresponding prediction model based on each asset interval training, the method further includes:

[0016] Generate new indicators based on business meaning and data time window and use them as feature inputs of the prediction model;

[0017] A technical method is used to preliminarily screen features to obtain preliminary features. The method of screening features is related to the selected model. If a logistic regression algorithm is used to train the model, any one of stability, concentration, IV value, correlation, multicollinearity, and significance is used to preliminarily screen features. If an integrated learning algorithm is used to train the model, any one of shap value and feature importance score is used to preliminarily screen features.

[0018] Among the preliminary features, features with strong business explanation capabilities are selected as feature inputs of the prediction model.

[0019] In some embodiments, the training of each asset interval to obtain a corresponding prediction model includes:

[0020] For each target variable 1, ..., k, a certain proportion of the target variable 0 is extracted to form k sets of training data;

[0021] Divide the training set and test set, and use historical data to train logistic regression and ensemble learning models respectively.

[0022] In some embodiments, the target customer is assigned to the asset interval with the maximum predicted probability, and the final predicted probability and asset improvement amount interval of the target customer are obtained, including:

[0023] The characteristic data of the target customers are respectively passed into the trained prediction model to obtain the predicted probability of the customer in each increase amount interval, wherein the trained prediction model integrates the prediction results of multiple models;

[0024] The customer is assigned to the increase amount range with the maximum predicted probability, and the customer asset increase probability and increase amount range are obtained.

[0025] In some embodiments, after assigning the target customer to the asset interval with the maximum predicted probability and obtaining the final predicted probability and asset improvement amount interval of the target customer, the method further includes:

[0026] Divide the predicted probability into multiple intervals, and count the model evaluation indicators and business indicators of each interval, wherein the business indicators include at least one of the following: the number of upgraded customers, the actual upgrade rate, the AUM increase amount, and the AUM increase rate;

[0027] Classify target customers into high-potential asset enhancement customers and low-potential asset enhancement customers according to the preset probability threshold;

[0028] According to the predicted asset enhancement amount, the high-potential asset enhancement customers are divided into multiple customer groups, and a corresponding marketing strategy is formulated for each customer group.

[0029] In some embodiments, dividing the target customer asset increase amount into a plurality of asset intervals, and obtaining a corresponding prediction model based on each asset interval training, includes:

[0030] According to the institutional business classification, the increase amount of target customer assets is divided into multiple intervals, and each interval is independently trained to obtain multiple prediction models, and the machine learning model with the best effect is selected as the target prediction model based on the model evaluation index.

[0031] In a second aspect, an embodiment of the present application further provides a client asset enhancement prediction device, which is executed by an asset management server, and the device includes:

[0032] A training module, used to divide the target customer asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training;

[0033] A first determination module, configured to obtain a predicted probability of a target customer in each asset interval according to each prediction model;

[0034] The second determination module is used to classify the target customer into the asset interval with the maximum prediction probability, and obtain the final prediction probability and asset improvement amount interval of the target customer.

[0035] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a processor; and a memory arranged to store computer executable instructions, wherein the executable instructions, when executed, cause the processor to perform the above method.

[0036] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple application programs, the electronic device executes the above method.

[0037] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: first, the target customer asset increase amount is divided into multiple asset intervals, and the corresponding prediction model is obtained based on the training of each asset interval. The prediction model is used to predict the probability value of each asset interval. Further, the predicted probability of the target customer in each asset interval is obtained according to each prediction model, and the target customer is attributed to the asset interval where the maximum prediction probability is located, so as to obtain the final prediction probability and asset increase amount interval of the target customer. The above method proposes to divide the customer asset increase amount into multiple intervals, and trains an independent prediction model for each interval, so as to achieve a more refined prediction. At the same time, by independently training multiple models, each model focuses on a specific increase amount interval, and can more deeply learn the customer characteristics and behavior patterns within the interval, further improving the prediction accuracy. Different from the problem of a single prediction dimension in the related art, the above method can solve it better.

[0038] In addition, it should be noted that by dividing the target customer's asset improvement amount into multiple asset intervals, a corresponding prediction model is obtained based on the training of each asset interval, and there is an intrinsic correlation between determining the target customer's prediction probability and the asset improvement interval to which the target user belongs. This intrinsic correlation conforms to the laws of nature, and based on this, the problem of how to identify potential asset improvement customers and improve the accuracy of identification is solved, and the corresponding technical effect is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0040] Figure 1 A schematic diagram of the process of predicting customer asset enhancement in an embodiment of the present application;

[0041] Figure 2 A schematic diagram of the logical relationship among target variable setting, model training, model application, and target customer group division in the customer asset enhancement prediction method in the embodiment of the present application;

[0042] Figure 3 This is a schematic diagram of feature screening of the customer asset enhancement prediction method in the embodiment of the present application;

[0043] Figure 4A schematic diagram of setting a probability threshold value of a method for predicting customer asset enhancement in an embodiment of the present application;

[0044] Figure 5 This is a schematic diagram of the structure of the customer asset improvement prediction device in the embodiment of the present application;

[0045] Figure 6 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0047] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0048] The present application embodiment provides a method for predicting customer asset improvement, such as Figure 1 As shown, a schematic diagram of the customer asset improvement prediction method in an embodiment of the present application is provided, and the method at least includes the following steps S110 to S130:

[0049] Step S110, dividing the target customer asset increase amount into a plurality of asset intervals, and obtaining a corresponding prediction model based on training of each asset interval.

[0050] Target customers refer to customers who need to predict the probability and amount of asset improvement, and the information of target customers will be stored in the asset management server in advance. The probability and amount of asset improvement of target customers need to be obtained through prediction. Specifically, the amount of asset improvement of target customers can be divided into multiple asset intervals. Preferably, a corresponding prediction model is obtained based on the training of each asset interval, and the probability of improvement of customers in each interval is predicted by dividing the amount of improvement into multiple intervals and training the model independently. This refined processing method makes the prediction results more detailed and specific, so as to more accurately reflect the possibility and amount range of customer asset improvement.

[0051] Furthermore, the prediction model is used to predict the probability of a customer's upgrade in each of the asset intervals, which can improve the prediction accuracy. By independently training multiple prediction models, each prediction model focuses on a specific upgrade amount interval, so that the customer characteristics and behavior patterns in the interval can be learned more deeply. This separate prediction strategy reduces the generalization error of the model in the global scope and improves the prediction accuracy in each interval.

[0052] Step S120, obtaining the predicted probability of the target customer in each of the asset intervals according to each of the prediction models.

[0053] According to the prediction results of each prediction model, the prediction probability of the target customer in each asset interval is obtained. It can be understood that since an independent prediction model is used for prediction for each asset interval, the prediction probabilities obtained for each asset interval are independent of each other.

[0054] Step S130, assigning the target customer to the asset interval with the maximum predicted probability, and obtaining the target customer's final predicted probability and asset improvement amount interval.

[0055] According to the obtained predicted probability of each asset interval, the target customer is assigned to the asset interval where the maximum predicted probability is located, and the maximum predicted probability and the asset interval where the maximum predicted probability is located are used as the final predicted probability and asset improvement amount interval of the target customer.

[0056] Different from most methods in related technologies that only focus on the probability of customer asset improvement, but ignore the prediction of the amount of improvement, and fail to meet the needs of financial institutions to formulate personalized service strategies and precision marketing. Through the above method, the target customer prediction probability and the asset improvement interval to which the target user belongs can be determined according to the prediction probability of each asset interval, and then, according to the asset improvement interval to which the target customer belongs, the customer asset improvement prediction result can be obtained.

[0057] Different from the manual prediction in the related art, different business personnel have different experiences, judgment criteria and risk preferences, which may lead to subjective and inconsistent prediction results. Through the above method, according to the probability value of each asset interval, the target customer prediction probability and the asset improvement interval to which the target user belongs are determined. Based on the objective and consistent asset improvement interval to which the target customer belongs, the customer asset improvement prediction result is obtained.

[0058] Through the above method, the prediction dimension is increased to a certain extent. The prediction of target customer asset increase is transformed from a single probability value to probability values ​​of multiple increase amount ranges. It not only predicts the probability of customer asset increase, but also further predicts the increase amount range to obtain the corresponding predicted increase amount range.

[0059] In one embodiment of the present application, before obtaining the corresponding prediction model based on each asset interval training, it also includes: obtaining information that affects the change of customer assets during a preset observation period for the target customer, wherein the information that affects the change of customer assets includes at least one of the following: basic information, attribute information, account information, asset information, transaction information, and preference information; dividing the increase amount of the target customer's assets into k asset intervals and setting target variables 1,..., k respectively, and setting the target variables of the target customer whose assets have not been increased to the above asset intervals to 0.

[0060] like Figure 2 As shown in the figure, for the predicted probability of the target customer's assets, it is necessary to first divide the increase amount into different increase amount intervals to obtain labels 0, 1, 2...k. Based on labels 0, 1, 2...k, where k is a natural number, models 1, model 2...model k are trained respectively. Based on the prediction results of multiple models, the customer is assigned to the increase amount interval with the maximum prediction probability, and low-potential increase customers and high-potential increase customers are distinguished. Finally, the customers are divided into different customer groups and marketing is carried out for the customer groups.

[0061] When establishing a prediction model for the amount of increase and the probability of increase, it is necessary to collect data in advance: obtain the target customer's basic information, attribute information, account information, asset information, transaction information, preference information, and other information that may affect the change of customer assets during the observation period. Then set the target variable: divide the amount of increase into k intervals, set the target variables 1,...,k respectively, and set the target variables of assets that are not increased to these k intervals to 0. Finally, perform data preprocessing: handle missing values ​​and outliers, process categorical variables through label encoding, one-hot encoding or other methods, and if the logistic regression algorithm is used to train the model, continuous variables can be binned.

[0062] It should be noted that the basic information includes, but is not limited to, the target user's identity information and unit information.

[0063] It should be noted that the attribute information includes, but is not limited to, transaction attributes and investment attributes related to the target user's assets.

[0064] It should be noted that account information includes, but is not limited to, personal financial assets such as customers' deposits in banks or financial institutions and various investment products purchased through banks or financial institutions. Investments mainly include funds, government bonds, insurance and investment products issued by financial institutions.

[0065] It should be noted that asset information includes but is not limited to financial management, savings and other asset information.

[0066] It should be noted that transaction information includes, but is not limited to, investment transaction information involving financial management, savings, etc.

[0067] It should be noted that preference information includes, but is not limited to, preferred financial products, preferred savings habits, etc.

[0068] like Figure 2 As shown in the figure, during the model training phase, data sampling is required before the model is trained. During data sampling, for each target variable 1, ..., k, a certain proportion of the target variable 0 is extracted to form k sets of training data. When training the model, it is necessary to divide the training set and the test set, and use historical data to train the logistic regression and ensemble learning models respectively.

[0069] In one embodiment of the present application, before obtaining the corresponding prediction model based on the training of each asset interval, it also includes: generating new indicators according to the business meaning and the data time window and using them as the feature input of the prediction model; using technical methods to preliminarily screen features to obtain preliminary features, and the method of screening features is related to the selected model. If a logistic regression algorithm is used to train the model, any one of stability, concentration, IV value, correlation, multicollinearity, and significance is used to preliminarily screen features. If an integrated learning algorithm is used to train the model, any one of shap value and feature importance score is used to preliminarily screen features; features with strong business explanation capabilities are screened out from the preliminary features as the feature input of the prediction model.

[0070] For feature derivation: derive new indicators based on business meaning and data time window. For example, if there is an indicator of the cumulative amount of money received in the month and an indicator of the cumulative amount of money paid out in the month, a new indicator can be derived: the ratio of money received and paid out in the month = the cumulative amount of money paid out in the month / the cumulative amount of money received in the month, and so on.

[0071] like Figure 3 As shown, for feature screening, technical methods are used to screen features: if the logistic regression algorithm is used to train the model, features can be initially screened by methods including but not limited to stability, concentration, iv value, correlation, multicollinearity, significance, etc. For feature screening, if an integrated learning algorithm is used to train the model, including but not limited to random forest, XGBooST, etc., features can be initially screened by methods including but not limited to shap value, feature importance score, etc.

[0072] Furthermore, features with strong business explanation capabilities are screened, and features with both technical prediction capabilities and good business explanations are finally retained. It can be understood that when interpreting the input features, features with strong business relevance need to be considered. The above-mentioned ensemble learning algorithm and logistic regression algorithm training model are only examples and are not used to limit the scope of protection of this application.

[0073] Through the above method, the increase amount is divided into multiple intervals, and the model is trained independently to predict the probability of the customer's increase in each interval. This refined processing method makes the prediction results more detailed and specific, and can more accurately reflect the possibility and amount range of customer asset increase, thereby achieving more refined predictions.

[0074] In one embodiment of the present application, the corresponding prediction model is obtained based on the training of each asset interval, including: for each target variable 1,...,k, a certain proportion of the target variable 0 is extracted to form k groups of training data; the training set and the test set are divided, and the logistic regression and ensemble learning models are trained respectively using historical data.

[0075] During the training phase of the model, the improvement amount is divided into k intervals, and the target variables are set to 1,...,k respectively. The target variables for assets that are not improved to these k intervals are set to 0.

[0076] Preferably, data preprocessing is also required to handle missing values ​​and outliers, and to process categorical variables through label encoding, one-hot encoding or other methods. In addition, if the model is trained using a logistic regression algorithm, continuous variables can also be binned.

[0077] In one embodiment of the present application, the target customer is assigned to the asset interval where the maximum prediction probability is located, and the final prediction probability and asset improvement amount interval of the target customer are obtained, including: the characteristic data of the target customer are respectively passed into the trained prediction model to obtain the prediction probability of the customer in each improvement amount interval, and the trained prediction model integrates the prediction results of multiple models; the customer is assigned to the improvement amount interval to which the maximum prediction probability belongs, and the customer's asset improvement probability and improvement amount interval are obtained.

[0078] The characteristic data of the target customers are respectively passed into the trained k potential improvement models to obtain the predicted probability of the customer in each improvement amount range, and the customer is assigned to the improvement amount range with the maximum predicted probability. In this way, the customer asset improvement probability and improvement amount range are obtained.

[0079] In one embodiment of the present application, after the target customer is assigned to the asset interval where the maximum predicted probability is located and the final predicted probability and asset improvement amount interval of the target customer are obtained, it also includes: dividing the predicted probability into multiple intervals, and counting the model evaluation indicators and business indicators of each interval, wherein the business indicators include at least one of the following: the number of upgraded customers, the actual upgrade rate, the Aum improvement amount, and the Aum improvement rate; dividing the target customers into high-potential asset improvement customers and low-potential asset improvement customers according to a preset probability threshold; dividing the high-potential asset improvement customers into multiple customer groups according to the predicted asset improvement amount, and formulating corresponding marketing strategies for each customer group.

[0080] like Figure 4 As shown in the figure, when setting the probability threshold: the predicted probability is divided into multiple intervals, and the model evaluation indicators and business indicators such as the number of upgraded customers, actual upgrade rate, AUM increase amount, and AUM increase rate in each interval are counted. The probability threshold is set to divide customers into high-potential and low-potential asset upgrade customers. Then, the target customer groups are divided, and the high-potential asset upgrade customers are divided into k customer groups according to the predicted asset increase amount, and corresponding marketing strategies are formulated for each customer group.

[0081] The AUM value (Asset Under Management) has become one of the key indicators for measuring a bank's retail high-net-worth customers. Literally, it refers to the scale of assets under management. Generally speaking, the higher the customer's AUM, the greater his contribution to the bank.

[0082] In one embodiment of the present application, the increase amount of the target customer's assets is divided into multiple asset intervals, and a corresponding prediction model is obtained based on the training of each asset interval, including: according to the institutional business classification, the increase amount of the target customer's assets is divided into multiple intervals, and each interval is independently trained to obtain multiple prediction models, and the machine learning model with the best effect is selected according to the model evaluation index as the target prediction model.

[0083] According to the results of different institutional business classifications, the target customer asset increase amount is divided into multiple intervals, and then each interval is independently trained to obtain multiple prediction models. Figure 2 As shown, it is also necessary to evaluate the model based on model indicators such as accuracy, recall, precision, auc, f1, lift, etc., and select the best machine learning model as the target prediction model.

[0084] Preferably, the target prediction model selection process also includes adjusting parameters to retrain the model and evaluating indicators until the performance of the model reaches a satisfactory (convergent) result.

[0085] The present application embodiment also provides a client asset enhancement prediction device 500, such as Figure 5As shown, a schematic diagram of the structure of the customer asset enhancement prediction device in an embodiment of the present application is provided, and the customer asset enhancement prediction device 500 at least includes: a training module 510, a first determination module 520, and a second determination module 530, wherein:

[0086] In one embodiment of the present application, the training module 510 is specifically used to divide the increase amount of the target customer's assets into multiple asset intervals, and obtain a corresponding prediction model based on the training of each asset interval.

[0087] Target customers refer to customers who need to predict the probability and amount of asset improvement, and the information of target customers will be stored in the asset management server in advance. The probability and amount of asset improvement of target customers need to be obtained through prediction. Specifically, the amount of asset improvement of target customers can be divided into multiple asset intervals. Preferably, a corresponding prediction model is obtained based on the training of each asset interval, and the probability of improvement of customers in each interval is predicted by dividing the amount of improvement into multiple intervals and training the model independently. This refined processing method makes the prediction results more detailed and specific, so as to more accurately reflect the possibility and amount range of customer asset improvement.

[0088] Furthermore, the prediction model is used to predict the probability of a customer's upgrade in each of the asset intervals, which can improve the prediction accuracy. By independently training multiple prediction models, each prediction model focuses on a specific upgrade amount interval, so that the customer characteristics and behavior patterns in the interval can be learned more deeply. This separate prediction strategy reduces the generalization error of the model in the global scope and improves the prediction accuracy in each interval.

[0089] In one embodiment of the present application, the first determination module 520 is specifically used to obtain the predicted probability of the target customer in each asset interval according to each prediction model.

[0090] According to the prediction results of each prediction model, the prediction probability of the target customer in each asset interval is obtained. It can be understood that since an independent prediction model is used for prediction for each asset interval, the prediction probabilities obtained for each asset interval are independent of each other.

[0091] In one embodiment of the present application, the second determination module 530 is specifically used to: assign the target customer to the asset interval where the maximum predicted probability is located, and obtain the final predicted probability and asset improvement amount interval of the target customer.

[0092] According to the obtained predicted probability of each asset interval, the target customer is assigned to the asset interval where the maximum predicted probability is located, and the maximum predicted probability and the asset interval where the maximum predicted probability is located are used as the final predicted probability and asset improvement amount interval of the target customer.

[0093] In one embodiment of the present application, it further includes: a preprocessing module for

[0094] Obtain information that affects changes in customer assets during a preset observation period, where the information that affects changes in customer assets includes at least one of the following: basic information, attribute information, account information, asset information, transaction information, and preference information; divide the increase amount of the target customer's assets into k asset intervals and set target variables 1, ..., k respectively, and set the target variables of the target customer's assets that are not increased to the above asset intervals to 0.

[0095] In one embodiment of the present application, the preprocessing module is also used to

[0096] According to the business meaning and the data time window, new indicators are generated and used as the feature input of the prediction model; technical methods are used to preliminarily screen features to obtain preliminary features. The method of screening features is related to the selected model. If a logistic regression algorithm is used to train the model, any one of stability, concentration, IV value, correlation, multicollinearity, and significance is used to preliminarily screen features. If an integrated learning algorithm is used to train the model, any one of shap value and feature importance score is used to preliminarily screen features; features with strong business explanation capabilities are screened out from the preliminary features as the feature input of the prediction model.

[0097] In one embodiment of the present application, the first determining module 520 is further configured to:

[0098] For each target variable 1, ..., k, a certain proportion of the target variable 0 is extracted to form k groups of training data; the training set and the test set are divided, and the historical data are used to train the logistic regression and ensemble learning models respectively.

[0099] In one embodiment of the present application, the second determining module 530 is further configured to:

[0100] The characteristic data of the target customers are respectively passed into the trained prediction models to obtain the predicted probability of the customer in each increase amount range. The trained prediction model integrates the prediction results of multiple models; the customer is assigned to the increase amount range to which the maximum prediction probability belongs, and the customer asset increase probability and increase amount range are obtained.

[0101] In one embodiment of the present application, the second determining module 630 is further configured to:

[0102] The predicted probability is divided into multiple intervals, and the model evaluation indicators and business indicators of each interval are counted, and the business indicators include at least one of the following: the number of upgraded customers, the actual upgrade rate, the Aum increase amount, and the Aum increase rate; the target customers are divided into high-potential asset improvement customers and low-potential asset improvement customers according to the preset probability threshold; according to the predicted asset increase amount, the high-potential asset improvement customers are divided into multiple customer groups, and corresponding marketing strategies are formulated for each customer group.

[0103] In one embodiment of the present application, the training module 610 is also used to

[0104] According to the institutional business classification, the increase amount of target customer assets is divided into multiple intervals, and each interval is independently trained to obtain multiple prediction models, and the machine learning model with the best effect is selected as the target prediction model based on the model evaluation index.

[0105] It can be understood that the above-mentioned customer asset enhancement prediction device can implement each step of the user asset prediction method provided in the aforementioned embodiment, and the relevant explanations about the user asset prediction method are applicable to the customer asset enhancement prediction device and will not be repeated here.

[0106] Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 6 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.

[0107] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0108] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0109] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a customer asset improvement prediction device at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations:

[0110] Divide the target customer's asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training;

[0111] Obtaining the predicted probability of the target customer in each asset range according to each of the prediction models;

[0112] Assign the target customer to the asset range with the maximum predicted probability, and obtain the target customer's final predicted probability and asset improvement amount range.

[0113] The above application Figure 1The method performed by the customer asset enhancement prediction device disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in software form. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0114] The electronic device may also perform Figure 1 The method for executing the customer asset enhancement prediction device in the Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0115] The present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by an electronic device including multiple application programs, enable the electronic device to execute Figure 1 The method performed by the customer asset enhancement prediction device in the illustrated embodiment is specifically used to perform:

[0116] Divide the target customer's asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training;

[0117] Obtaining the predicted probability of the target customer in each asset range according to each of the prediction models;

[0118] Assign the target customer to the asset range with the maximum predicted probability, and obtain the target customer's final predicted probability and asset improvement amount range.

[0119] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0121] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0123] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0124] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0125] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0126] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0128] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for predicting customer asset enhancement, wherein: Executed by the asset management server, the prediction method includes: Divide the target customer's asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training; Obtaining the predicted probability of the target customer in each asset range according to each of the prediction models; Assign the target customer to the asset range with the maximum predicted probability, and obtain the target customer's final predicted probability and asset improvement amount range.

2. The method of claim 1, wherein: Before obtaining the corresponding prediction model based on each asset interval training, the method further includes: Acquire information that affects changes in the target customer's assets during a preset observation period, where the information that affects changes in the customer's assets includes at least one of the following: basic information, attribute information, account information, asset information, transaction information, and preference information; The increase amount of the target customer's assets is divided into k asset intervals and target variables 1, ..., k are set respectively. Meanwhile, the target variables of the target customer's assets that are not increased to the above asset intervals are set to 0.

3. The method of claim 1, wherein: Before obtaining the corresponding prediction model based on each asset interval training, the method further includes: Generate new indicators based on business meaning and data time window and use them as feature inputs of the prediction model; A technical method is used to preliminarily screen features to obtain preliminary features. The method of screening features is related to the selected model. If a logistic regression algorithm is used to train the model, any one of stability, concentration, IV value, correlation, multicollinearity, and significance is used to preliminarily screen features. If an integrated learning algorithm is used to train the model, any one of shap value and feature importance score is used to preliminarily screen features. Among the preliminary features, features with strong business explanation capabilities are selected as feature inputs of the prediction model.

4. The method of claim 1, wherein: The training based on each asset interval to obtain a corresponding prediction model includes: For each target variable 1, ..., k, a certain proportion of the target variable 0 is extracted to form k sets of training data; Divide the training set and test set, and use historical data to train logistic regression and ensemble learning models respectively.

5. The method of claim 4, wherein: Assign the target customer to the asset range with the maximum predicted probability, and obtain the target customer's final predicted probability and asset improvement amount range, including: The characteristic data of the target customers are respectively passed into the trained prediction model to obtain the predicted probability of the customer in each increase amount interval, wherein the trained prediction model integrates the prediction results of multiple models; The customer is assigned to the increase amount range with the maximum predicted probability, and the customer asset increase probability and increase amount range are obtained.

6. The method of claim 1, wherein: After assigning the target customer to the asset range with the maximum predicted probability and obtaining the target customer's final predicted probability and asset improvement amount range, it also includes: Divide the predicted probability into multiple intervals, and count the model evaluation indicators and business indicators of each interval, wherein the business indicators include at least one of the following: the number of upgraded customers, the actual upgrade rate, the AUM increase amount, and the AUM increase rate; Classify target customers into high-potential asset enhancement customers and low-potential asset enhancement customers according to the preset probability threshold; According to the predicted asset enhancement amount, the high-potential asset enhancement customers are divided into multiple customer groups, and a corresponding marketing strategy is formulated for each customer group.

7. The method of claim 1, wherein: The step of dividing the target customer asset increase amount into a plurality of asset intervals and obtaining a corresponding prediction model based on each asset interval training includes: According to the institutional business classification, the increase amount of target customer assets is divided into multiple intervals, and each interval is independently trained to obtain multiple prediction models, and the machine learning model with the best effect is selected as the target prediction model based on the model evaluation index.

8. A customer asset enhancement prediction device, wherein: Executed by an asset management server, the device includes: A training module, used to divide the target customer asset increase amount into multiple asset intervals, and obtain a corresponding prediction model based on each asset interval training; A first determination module, configured to obtain a predicted probability of a target customer in each asset interval according to each prediction model; The second determination module is used to classify the target customer into the asset interval with the maximum prediction probability, and obtain the final prediction probability and asset improvement amount interval of the target customer.

9. An electronic device, comprising: Processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, causes the electronic device to execute any one of the methods of claims 1 to 7.

Images