Potential customer fund prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses a potential customer fund prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining user consumption information which comprises the consumption flow information of a user in a specified financial institution; the user consumption information is input into a pre-trained prediction model, prediction output is obtained, the pre-trained prediction model is obtained through machine learning training by using multiple sets of data, and each set of data in the multiple sets of data comprises user consumption data and a label corresponding to the user consumption data; and obtaining a potential customer fund prediction result according to the prediction output. On one hand, a mode of predicting potential customer funds is provided, and on the other hand, the prediction accuracy of inactive users is improved.

Description

Technical Field

[0001] The present application relates to the technical field of financial data analysis, and in particular to a method, device, electronic device, and storage medium for predicting funds of potential customers. Background Art

[0002] In the banking and financial industries, when conducting customer management, account managers usually need to recommend matching products based on the client's asset capacity to improve customer stickiness and customer experience. However, since customers usually do not place all their assets in the same institution, account managers are unable to correctly assess the client's actual assets, and thus misjudge the client's asset level and recommend incompatible products.

[0003] Recommendations are limited by the products provided by the financial institution. If the products provided by the financial institution cannot attract customers and the customers do not make purchases, the customer value cannot be accurately estimated. Furthermore, recommendations are also limited by the customer's investment habits and user stickiness. Due to sparse data, inactive customers cannot be accurately estimated. Summary of the invention

[0004] The embodiments of the present application provide a potential customer fund prediction method, device, electronic device, and storage medium to assist account managers in more accurately determining the customer's asset level and providing corresponding product recommendations.

[0005] The present application embodiment adopts the following technical solutions:

[0006] In a first aspect, an embodiment of the present application provides a potential customer funds prediction method, wherein the method comprises:

[0007] Obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution;

[0008] Inputting the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each set of data in the multiple sets of data includes: user consumption data and a label corresponding to the user consumption data;

[0009] According to the prediction output, the result of potential customer funds prediction is obtained.

[0010] In some embodiments, the pre-trained prediction model includes:

[0011] Obtaining the user's consumption flow information at a designated financial institution, wherein the consumption flow information includes at least one of the following: consumption location, consumption amount, product type, and brand data.

[0012] Labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at the designated financial institution to obtain label data;

[0013] A pre-trained prediction model is obtained by training based on the consumption flow information of the user at the designated financial institution and the label data corresponding to the consumption flow information of the user at the designated financial institution.

[0014] In some embodiments, the step of labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at a designated financial institution to obtain label data includes:

[0015] Divide the consumption amount into corresponding consumption levels;

[0016] The brand data is jointly determined by the transaction merchant name and the third-party interface disclosed by the e-commerce company;

[0017] Determine the product type according to the merchant type code;

[0018] Determine a brand price range based on the brand data and the product type;

[0019] According to the brand price range and the corresponding consumption level, the consumption levels are divided to obtain consumption value labels.

[0020] In some embodiments, the step of labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at a designated financial institution to obtain label data further includes:

[0021] Label the consumption channel preference according to the payment channel and payment object to obtain the consumption channel label;

[0022] Label the consumption locations, frequencies, average amounts, and maximum amounts of each type of consumption by customers to obtain labels for consumption and activity locations;

[0023] Aggregate consumption channels and consumption value labels according to different time periods to obtain consumption frequency and amount labels;

[0024] The customer’s average annual AUM within a preset period of time is used as the initial value for normalization, and the customer value coefficient is obtained as the customer value label.

[0025] In some embodiments, obtaining a result of potential customer funds prediction based on the prediction output includes:

[0026] Using the users with the most account activation times in the designated financial institution as customer consumption training data, and obtaining customer value labels according to labels corresponding to the customer consumption training data;

[0027] The customer consumption training data is used as a feature and the customer value label is used as a prediction value to train a machine learning model to predict and output a predicted consumption value coefficient;

[0028] The customer asset level is fitted according to the predicted consumption value coefficient, and the consumption value coefficient and the customer's actual asset level are fitted, and then the predicted value of the customer value is obtained after normalization processing.

[0029] In some embodiments, before inputting the user consumption information into a pre-trained prediction model, the process includes:

[0030] The consumption frequency in the user consumption information is modified according to the customer stickiness while the single consumption amount and the maximum consumption amount in the user consumption information are not modified. The lower the customer stickiness, the lower the stickiness. The method for obtaining the customer stickiness index includes:

[0031] Obtain the historical behavior data of the user in the designated financial institution, and perform statistical sorting of all customer data for each item of the historical behavior data;

[0032] Based on the results of the ranking, each indicator is scored using quantiles;

[0033] Based on the scoring results of the various indicators, the indicators are scored and calculated in a weighted average manner and normalized to obtain the customer stickiness indicator.

[0034] In some embodiments, the modifying the consumption frequency in the user consumption information according to the customer stickiness includes:

[0035] Based on various types of consumption and various levels of consumption, the median of customers who meet the preset loyalty value is determined to obtain the corrected maximum value;

[0036] According to the actual consumption times and the customer stickiness, a revised consumption times is obtained;

[0037] If the revised consumption amount is higher than the revised maximum value and the original value is lower than the maximum value, the revised maximum value is taken;

[0038] Otherwise, take the corrected consumption times.

[0039] In a second aspect, an embodiment of the present application further provides a potential customer funds prediction device, wherein the device comprises:

[0040] An acquisition module, used to acquire user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution;

[0041] A prediction module, used to input the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each of the multiple sets of data including: user consumption data and a label corresponding to the user consumption data;

[0042] The output module is used to obtain the result of potential customer fund prediction based on the prediction output.

[0043] 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.

[0044] 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.

[0045] At least one of the above-mentioned technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information in a designated financial institution. By inputting the user consumption information into a pre-trained prediction model, a prediction output is obtained. The pre-trained prediction model is obtained by machine learning training using multiple sets of data, and each set of data in the multiple sets of data includes: user consumption data and labels corresponding to the user consumption data. Finally, based on the prediction output, the result of the potential customer fund prediction is obtained. Through the above method, inactive customers can be predicted, and it does not rely on the customer's financial situation and financial product situation in the designated institution, and only sparse consumption characteristics are used to predict the customer's asset hierarchy level. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] 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:

[0047] Figure 1 This is a flow chart of a potential customer funds prediction method in an embodiment of the present application;

[0048] Figure 2 This is a schematic diagram of the structure of a potential customer funds prediction device in an embodiment of the present application;

[0049] Figure 3 This is a schematic diagram of the structure of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0050] 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.

[0051] The technical solutions provided by various embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0052] The present application embodiment provides a potential customer funds prediction method, such as Figure 1 As shown, a flow chart of a potential customer fund prediction method in an embodiment of the present application is provided, and the potential customer fund prediction method at least includes the following steps S110 to S130:

[0053] Step S110, obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution.

[0054] In the data preparation stage, the data object for modeling is consumption data. Based on the consumption data, the customer's consumption habits and consumption capacity are portrayed, and his asset capacity is inferred.

[0055] Considering that the consumption data obtained by banks or financial institutions is more sparse than that obtained through e-commerce and other channels, the main information used for user consumption is the consumption amount, transaction counterparty (merchant), transaction time, etc. in the customer's consumption flow.

[0056] It should be noted that user consumption information mainly includes the user's consumption flow information at a designated financial institution, and the user in the embodiment of the present application is a user of a designated financial institution. Therefore, the consumption flow information of the designated financial institution can be obtained, but the consumption flow information of other financial institutions cannot be obtained.

[0057] Step S120, inputting the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning training using multiple sets of data, and each set of data in the multiple sets of data includes: user consumption data and a label corresponding to the user consumption data.

[0058] Consumption behavior mainly includes multiple types of data as training data: consumption location, consumption product type, consumption frequency, product brand, consumption amount, etc. Because only part of the customer's transaction information can be obtained, the consumption location, product type and brand data are mainly selected as modeling data, and consumption frequency, account activity frequency, etc. can be selected as activity indicators to fit the penalty function to correct the modeling input features.

[0059] Considering that the funds of potential customers depend on financial payment habits, they are not completely independent of consumption. However, in the embodiments of the present application, it is intended to reflect the real consumption level of the target user through consumption. For example, some users are accustomed to using a card of a certain financial institution for consumption. For another example, different customers have different consumption habits. Some like long-term large-amount consumption, while others like small-amount short-term consumption. If only according to the consumption frequency, it is impossible to objectively reflect the real consumption level of the target user.

[0060] The training data of the model includes user consumption data and labels corresponding to the user consumption data (including but not limited to consumption channel labels, consumption and activity venue labels, frequency and amount labels). In the embodiments of the present application, asset labeling is enriched and user consumption is segmented in order to predict more accurate results. These labels are used as model input variables to predict results.

[0061] Step S130, obtaining a result of potential customer fund prediction based on the prediction output.

[0062] The prediction output results need to be fitted with other results before the final result of the potential customer fund prediction is obtained. It can be understood that the prediction of potential customer funds is not a simple financial problem, but a result obtained after analyzing and processing financial data that conforms to objective natural laws.

[0063] The results of potential customer fund prediction are pre-processed based on the full data of user consumption information, and dormant users are eliminated. The data is then corrected using the corresponding input feature correction method, and the consumption value is predicted using a pre-trained prediction model to further infer actual assets.

[0064] Through the above method, the potential customer funds can be accurately predicted. In particular, for inactive customers, the consumption characteristics can be used to accurately predict the customer asset level.

[0065] Through the above method, the labeling process in the user consumption information is optimized, so that the actual consumption level of the customer is reflected more objectively.

[0066] In one embodiment of the present application, the pre-trained prediction model includes: obtaining the consumption flow information of the user at a designated financial institution, the consumption flow information including at least one of the following: consumption location, consumption amount, product type, and brand data, labeling the consumption location, consumption amount, product type, and brand data in the consumption flow information of the user at the designated financial institution to obtain label data; training based on the consumption flow information of the user at the designated financial institution and the label data corresponding to the consumption flow information of the user at the designated financial institution to obtain the pre-trained prediction model.

[0067] The consumption location refers to the place where users make purchases through banks or financial institutions, for example, a shopping mall in Haidian District, Beijing.

[0068] The consumption amount refers to the actual amount of consumption by the user each time.

[0069] Product type refers to the type of product purchased by the user.

[0070] Brand data refers to data related to the brands purchased by users.

[0071] Based on the above consumption flow information, the consumption location, consumption amount, product type and brand data in the consumption flow information of the user in the designated financial institution are marked to obtain label data. After marking, the consumption location, consumption amount, product type and brand data will be finely divided.

[0072] Based on the above consumption flow information, according to the consumption flow information of the user in the designated financial institution and the label data corresponding to the consumption flow information of the user in the designated financial institution, a pre-trained prediction model is trained. According to the consumption flow information of the designated financial institution and the corresponding label data, the trained prediction model can predict the funds of potential customers. The user's consumption information in other financial institutions or other consumption information is not required.

[0073] In one embodiment of the present application, the consumption location, the consumption amount, the product type and the brand data in the consumption flow information of the user in a designated financial institution are labeled to obtain label data, including: dividing the consumption amount into corresponding consumption levels; jointly determining the brand data through the transaction merchant name and the third-party interface disclosed by the e-commerce company; determining the product type according to the merchant type code; determining the brand price range according to the brand data and the product type; and dividing the consumption levels according to the brand price range and the corresponding consumption levels to obtain a consumption value label.

[0074] First, consumer behavior is divided into ultra-high, high, medium, and low-end consumption.

[0075] Secondly, the merchant brand is determined by the transaction merchant name and the e-commerce public API interface, and the product type is determined by the merchant type code.

[0076] Then, after the brand extraction is completed, the brand price range is determined based on the ranking of single consumption within each product type and the price of this brand on the e-commerce platform, and the consumption levels are divided.

[0077] Finally, calculate the product type and merchant brand combination for consumption value labeling.

[0078] In one embodiment of the present application, the consumption location, the consumption amount, the product type and the brand data in the consumption flow information of the user in the designated financial institution are labeled to obtain label data, and also includes: labeling consumption channel preferences according to payment channels and payment objects to obtain consumption channel labels; labeling the consumption places, frequencies, average amounts and maximum amounts of various types of consumption of customers to obtain consumption and activity place labels; aggregating on consumption channels and consumption value labels according to different time periods to obtain consumption frequency and amount labels; and normalizing according to the customer's average annual AUM within a preset time as the initial value to obtain a customer value coefficient as a customer value label.

[0079] The consumption channel preference is marked according to the payment channel and the payment object to obtain the consumption channel label.

[0080] Label the customer dimension and obtain labels for consumption and activity venues based on the consumption locations, frequency, average amount, and maximum amount of each type of consumption by the customer.

[0081] The consumption frequency and amount are aggregated based on the consumption channel and consumption value labels for the current month, last month, the past three months, and the past six months to obtain the consumption frequency and amount labels.

[0082] When training the model, the above customer consumption data is used as features.

[0083] In addition, the customer's average annual AUM in the past year is used as the initial value for normalization. Z-score normalization can be selected to obtain the customer value coefficient. The formula for Z-score normalization is: The customer value label is obtained and used as a prediction value during model training.

[0084] Z-score normalization is a commonly used data normalization method, also known as standardization. It can transform data into a distribution with the same mean and standard deviation. The specific operation of Z-Score normalization is to subtract the mean μ of the data set from each data point x and then divide it by the standard deviation σ. In the field of machine learning, Z-Score normalization is often used in data processing, such as neural networks, cluster analysis and other applications.

[0085] In one embodiment of the present application, the result of the potential customer fund prediction is obtained based on the prediction output, including: taking the part of users who have the most account activation times in the designated financial institution as customer consumption training data, and obtaining customer value labels according to the labels corresponding to the customer consumption training data; using the customer consumption training data as features and the customer value labels as prediction values ​​to train a machine learning model for predicting and outputting a predicted consumption value coefficient; fitting the customer asset level according to the predicted consumption value coefficient, fitting according to the consumption value coefficient and the customer's actual asset level, and then obtaining the predicted value of the customer value after normalization.

[0086] The prediction output includes the process of fitting the prediction model, and the prediction results output by the prediction model also need to be optimized and fitted.

[0087] For example, the top 1% of customers with the most account activation times in all financial institutions or banks can be selected as training data, and consumption value scores can be performed. Then, customer value can be derived based on the consumption value scores. The training process is as follows, and the training of XGboost is used as an example for explanation:

[0088] First, the customer consumption data is used as features and the customer value label as the predicted value to train the XGboost model to predict the consumption value coefficient.

[0089] Secondly, the customer's asset level is fitted with the customer's predicted consumption value coefficient obtained according to the model prediction, and the regression curve is fitted with the consumption value coefficient and the customer's actual asset level. The calculation is performed using the Z-score normalization method, that is, X=Z·σ+μ, to obtain the corresponding predicted value of customer value.

[0090] Z-score normalization is a commonly used data normalization method, also known as standardization. It can transform data into a distribution with the same mean and standard deviation. The specific operation of Z-Score normalization is to subtract the mean μ of the data set from each data point x and then divide it by the standard deviation σ. In the field of machine learning, Z-Score normalization is often used in data processing, such as neural networks, cluster analysis and other applications.

[0091] It can be understood that, based on the normal distribution of potential customer funds, the predicted values ​​correspond to the amount of assets the potential customer has and the proportion of the assets.

[0092] It can be understood that the above 1% ratio can be configured according to the actual scenario and is not specifically limited in the embodiments of the present application.

[0093] It should be noted that the above XGboost is only an example and is not intended to limit the solutions in the embodiments of the present application.

[0094] In one embodiment of the present application, before inputting the user consumption information into a pre-trained prediction model, it includes: correcting the consumption frequency in the user consumption information according to the customer stickiness while not correcting the single consumption amount and the maximum consumption amount in the user consumption information. The lower the customer stickiness, the lower the stickiness. The method for obtaining the customer stickiness index includes: obtaining the user's historical behavior data in a designated financial institution, and performing statistical sorting of all customer data for each item of the historical behavior data; according to the result of the sorting, scoring each indicator using quantiles; based on the result of the scoring of each indicator, scoring and normalizing each indicator in a weighted average manner to obtain the customer stickiness index.

[0095] The prediction features are modified according to customer stickiness. By calculating the customer stickiness coefficient, the lower the coefficient, the lower the stickiness, the frequency of customer consumption on different consumer goods is modified, but the single consumption amount and the maximum consumption amount are not modified to deal with the situation where the customer does not mainly use the account of this institution as the main account.

[0096] The consumption frequency in the user consumption information is corrected according to the customer stickiness, while the single consumption amount and the maximum consumption amount in the user consumption information are not corrected. It should be noted that the lower the customer stickiness, the lower the stickiness, that is, low stickiness indicates low customer stickiness.

[0097] For example, consumption frequency, account activity frequency, etc. are selected as activity indicators to fit the penalty function and correct the modeling input features.

[0098] Specifically, the customer stickiness coefficient is calculated as follows:

[0099] Based on the number of product types held by customers, customer account age, number of account changes, number of consumption, number of transfers with the same name / different name, number of mobile banking logins, number of consecutive login days, benefits collection, benefits usage, etc., all customer data are statistically sorted and each indicator is scored using quantiles. Each indicator is then scored and calculated as a weighted average, and after min-max normalization, the customer stickiness indicator is obtained.

[0100] In one embodiment of the present application, the consumption frequency in the user consumption information is corrected according to customer stickiness, including: based on various types of consumption and various levels of consumption, determining the median of customers who meet the preset loyalty value to obtain a corrected maximum value; obtaining a corrected number of consumptions according to the actual number of consumptions and the customer stickiness; if the corrected consumption amount is higher than the corrected maximum value and the original value is less than the maximum value, taking the corrected maximum value; otherwise, taking the corrected number of consumptions.

[0101] For example, consumption frequency, account activity frequency, etc. are selected as activity indicators to fit the penalty function and correct the modeling input features.

[0102] To obtain better prediction results, the feature data of the input prediction model is optimized. The customer stickiness index corrects the customer's consumption frequency, takes the median of customers with a loyalty of more than 90 points for each type of consumption and each level of consumption, and takes it as the maximum value of the correction value, and then calculates the actual consumption frequency / cos (customer stickiness) as the corrected consumption frequency.

[0103] Case 1: If the revised amount is higher than the maximum value and the original value is lower than the maximum value, the maximum value is taken.

[0104] Case 2: Otherwise, take the corrected number of times, that is, when the corrected threshold is very large, remove the maximum value and take the original value.

[0105] It can be understood that a loyalty score of 90 points can represent the actual consumption coefficient of the most loyal users, and this can be used as the limit threshold.

[0106] The present application embodiment also provides a potential customer funds prediction device 200, such as Figure 2 As shown, a schematic diagram of the structure of a potential customer fund prediction device in an embodiment of the present application is provided, wherein the potential customer fund prediction device 200 at least includes: an acquisition module 210, a prediction module 220, and an output module 230, wherein:

[0107] In one embodiment of the present application, the acquisition module 210 is specifically used to: acquire user consumption information, where the user consumption information includes the user's consumption flow information at a designated financial institution.

[0108] In the data preparation stage, the data object for modeling is consumption data. Based on the consumption data, the customer's consumption habits and consumption capacity are portrayed, and his asset capacity is inferred.

[0109] Considering that the consumption data obtained by banks or financial institutions is more sparse than that obtained through e-commerce and other channels, the main information used for user consumption is the consumption amount, transaction counterparty (merchant), transaction time, etc. in the customer's consumption flow.

[0110] It should be noted that user consumption information mainly includes the user's consumption flow information at a designated financial institution, and the user in the embodiment of the present application is a user of a designated financial institution. Therefore, the consumption flow information of the designated financial institution can be obtained, but the consumption flow information of other financial institutions cannot be obtained.

[0111] In one embodiment of the present application, the prediction module 220 is specifically used to: input the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning training using multiple sets of data, and each set of data in the multiple sets of data includes: user consumption data and labels corresponding to the user consumption data.

[0112] Consumption behavior mainly includes multiple types of data as training data: consumption location, consumption product type, consumption frequency, product brand, consumption amount, etc. Because only part of the customer's transaction information can be obtained, the consumption location, product type and brand data are mainly selected as modeling data, and consumption frequency, account activity frequency, etc. can be selected as activity indicators to fit the penalty function to correct the modeling input features.

[0113] Considering that the funds of potential customers depend on financial payment habits, they are not completely independent of consumption. However, in the embodiments of the present application, it is intended to reflect the real consumption level of the target user through consumption. For example, some users are accustomed to using a card of a certain financial institution for consumption. For another example, different customers have different consumption habits. Some like long-term large-amount consumption, while others like small-amount short-term consumption. If only according to the consumption frequency, it is impossible to objectively reflect the real consumption level of the target user.

[0114] The training data of the model includes user consumption data and labels corresponding to the user consumption data (including but not limited to consumption channel labels, consumption and activity venue labels, frequency and amount labels). In the embodiments of the present application, asset labeling is enriched and user consumption is segmented in order to predict more accurate results. These labels are used as model input variables to predict results.

[0115] In one embodiment of the present application, the output module 230 is specifically used to obtain the result of potential customer fund prediction based on the prediction output.

[0116] The prediction output results need to be fitted with other results before the final result of the potential customer fund prediction is obtained. It can be understood that the prediction of potential customer funds is not a simple financial problem, but a result obtained after analyzing and processing financial data that conforms to objective natural laws.

[0117] The results of potential customer fund prediction are pre-processed based on the full data of user consumption information, and dormant users are eliminated. The data is then corrected using the corresponding input feature correction method, and the consumption value is predicted using a pre-trained prediction model to further infer actual assets.

[0118] In one embodiment of the present application, the prediction module 210 is also used to

[0119] Obtaining the user's consumption flow information at a designated financial institution, wherein the consumption flow information includes at least one of the following: consumption location, consumption amount, product type, and brand data.

[0120] Labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at the designated financial institution to obtain label data;

[0121] A pre-trained prediction model is obtained by training based on the consumption flow information of the user at the designated financial institution and the label data corresponding to the consumption flow information of the user at the designated financial institution.

[0122] In one embodiment of the present application, the prediction module 210 is also used to

[0123] Divide the consumption amount into corresponding consumption levels;

[0124] The brand data is jointly determined by the transaction merchant name and the third-party interface disclosed by the e-commerce company;

[0125] Determine the product type according to the merchant type code;

[0126] Determine a brand price range based on the brand data and the product type;

[0127] According to the brand price range and the corresponding consumption level, the consumption levels are divided to obtain consumption value labels.

[0128] In one embodiment of the present application, the prediction module 210 is also used to

[0129] Label the consumption channel preference according to the payment channel and payment object to obtain the consumption channel label;

[0130] Label the consumption locations, frequencies, average amounts, and maximum amounts of each type of consumption by customers to obtain labels for consumption and activity locations;

[0131] Aggregate consumption channels and consumption value labels according to different time periods to obtain consumption frequency and amount labels;

[0132] The customer’s average annual AUM within a preset period of time is used as the initial value for normalization, and the customer value coefficient is obtained as the customer value label.

[0133] In one embodiment of the present application, the output module 230 is also used to

[0134] Using the users with the most account activation times in the designated financial institution as customer consumption training data, and obtaining customer value labels according to labels corresponding to the customer consumption training data;

[0135] The customer consumption training data is used as a feature and the customer value label is used as a prediction value to train a machine learning model to predict and output a predicted consumption value coefficient;

[0136] The customer asset level is fitted according to the predicted consumption value coefficient, and the consumption value coefficient and the customer's actual asset level are fitted, and then the predicted value of the customer value is obtained after normalization processing.

[0137] In one embodiment of the present application, it also includes: an optimization module for

[0138] The consumption frequency in the user consumption information is modified according to the customer stickiness while the single consumption amount and the maximum consumption amount in the user consumption information are not modified. The lower the customer stickiness, the lower the stickiness. The method for obtaining the customer stickiness index includes:

[0139] Obtain the historical behavior data of the user in the designated financial institution, and perform statistical sorting of all customer data for each item of the historical behavior data;

[0140] Based on the results of the ranking, each indicator is scored using quantiles;

[0141] Based on the scoring results of the various indicators, the indicators are scored and calculated in a weighted average manner and normalized to obtain the customer stickiness indicator.

[0142] In one embodiment of the present application, the optimization module is also used to

[0143] Based on various types of consumption and various levels of consumption, the median of customers who meet the preset loyalty value is determined to obtain the corrected maximum value;

[0144] According to the actual consumption times and the customer stickiness, a revised consumption times is obtained;

[0145] If the revised consumption amount is higher than the revised maximum value and the original value is lower than the maximum value, the revised maximum value is taken;

[0146] Otherwise, take the corrected consumption times.

[0147] It can be understood that the above-mentioned potential customer funds prediction device can implement the various steps of the potential customer funds prediction method provided in the aforementioned embodiment, and the relevant explanations about the potential customer funds prediction method are applicable to the potential customer funds prediction device and will not be repeated here.

[0148] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 3 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.

[0149] 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 3 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.

[0150] 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.

[0151] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a potential customer fund prediction device at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations:

[0152] Obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution;

[0153] Inputting the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each set of data in the multiple sets of data includes: user consumption data and a label corresponding to the user consumption data;

[0154] According to the prediction output, the result of potential customer funds prediction is obtained.

[0155] The above application Figure 1 The method performed by the potential customer funds 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 the form of software. 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.

[0156] The electronic device may also perform Figure 1 The method is implemented by the potential customer funds prediction device in Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0157] 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 potential customer funds prediction device in the illustrated embodiment is specifically used to perform:

[0158] Obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution;

[0159] Inputting the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each set of data in the multiple sets of data includes: user consumption data and a label corresponding to the user consumption data;

[0160] According to the prediction output, the result of potential customer funds prediction is obtained.

[0161] 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.

[0162] 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.

[0163] 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 1A function specified in one or more boxes.

[0164] 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.

[0165] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0166] 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.

[0167] 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.

[0168] 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.

[0169] 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.

[0170] 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 funds of potential customers, wherein: The method comprises: Obtaining user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution; Inputting the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each set of data in the multiple sets of data includes: user consumption data and a label corresponding to the user consumption data; According to the prediction output, the result of potential customer funds prediction is obtained.

2. The method of claim 1, wherein: The pre-trained prediction model includes: Obtaining the user's consumption flow information at a designated financial institution, wherein the consumption flow information includes at least one of the following: consumption location, consumption amount, product type, and brand data. Labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at the designated financial institution to obtain label data; A pre-trained prediction model is obtained by training based on the consumption flow information of the user at the designated financial institution and the label data corresponding to the consumption flow information of the user at the designated financial institution.

3. The method of claim 2, wherein: The step of labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at the designated financial institution to obtain label data includes: Divide the consumption amount into corresponding consumption levels; The brand data is jointly determined by the transaction merchant name and the third-party interface disclosed by the e-commerce company; Determine the product type according to the merchant type code; Determine a brand price range based on the brand data and the product type; According to the brand price range and the corresponding consumption level, the consumption levels are divided to obtain consumption value labels.

4. The method of claim 2, wherein: The step of labeling the consumption location, the consumption amount, the product type, and the brand data in the consumption flow information of the user at the designated financial institution to obtain label data further includes: Label the consumption channel preference according to the payment channel and payment object to obtain the consumption channel label; Label the consumption locations, frequencies, average amounts, and maximum amounts of each type of consumption by customers to obtain labels for consumption and activity locations; Aggregate consumption channels and consumption value labels according to different time periods to obtain consumption frequency and amount labels; The customer’s average annual AUM within a preset period of time is used as the initial value for normalization, and the customer value coefficient is obtained as the customer value label.

5. The method of claim 4, wherein: The method of obtaining a result of potential customer fund prediction based on the prediction output includes: Using the users with the most account activation times in the designated financial institution as customer consumption training data, and obtaining customer value labels according to labels corresponding to the customer consumption training data; The customer consumption training data is used as a feature and the customer value label is used as a prediction value to train a machine learning model to predict and output a predicted consumption value coefficient; The customer asset level is fitted according to the predicted consumption value coefficient, and the consumption value coefficient and the customer's actual asset level are fitted, and then the predicted value of the customer value is obtained after normalization processing.

6. The method of claim 1, wherein: Before inputting the user consumption information into the pre-trained prediction model, the method includes: The consumption frequency in the user consumption information is modified according to the customer stickiness while the single consumption amount and the maximum consumption amount in the user consumption information are not modified. The lower the customer stickiness, the lower the stickiness. The method for obtaining the customer stickiness index includes: Obtain the historical behavior data of the user in the designated financial institution, and perform statistical sorting of all customer data for each item of the historical behavior data; Based on the results of the ranking, each indicator is scored using quantiles; Based on the scoring results of the various indicators, the indicators are scored and calculated in a weighted average manner and normalized to obtain the customer stickiness indicator.

7. The method of claim 6, wherein: The modifying of the consumption frequency in the user consumption information according to the customer stickiness includes: Based on various types of consumption and various levels of consumption, the median of customers who meet the preset loyalty value is determined to obtain the corrected maximum value; According to the actual consumption times and the customer stickiness, a revised consumption times is obtained; If the revised consumption amount is higher than the revised maximum value and the original value is lower than the maximum value, the revised maximum value is taken; Otherwise, take the corrected consumption times.

8. A potential customer funds prediction device, wherein: The device comprises: An acquisition module, used to acquire user consumption information, wherein the user consumption information includes the user's consumption flow information at a designated financial institution; A prediction module, used to input the user consumption information into a pre-trained prediction model to obtain a prediction output, wherein the pre-trained prediction model is obtained by machine learning using multiple sets of data, each of the multiple sets of data including: user consumption data and a label corresponding to the user consumption data; The output module is used to obtain the result of potential customer fund prediction based on the prediction output.

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