Financial product recommendation method and device based on interest rate prediction

Abstract

The present invention provides a method and device for recommending financial products based on interest rate prediction, which are suitable for the financial field. The method comprises: obtaining, based on financial product data and corresponding historical purchase record data, the support degree of two financial products purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product and then purchased another financial product, and the improvement degree of the application of association rules of the two financial products; obtaining a bank interest rate prediction curve based on a pre-trained long short-term memory neural network; calculating the recommendation degree of each group of financial products based on the financial product yield curve, the bank interest rate prediction curve and the improvement degree; and recommending financial products based on the support degree, the confidence degree and the recommendation degree, wherein the bank interest rate prediction curve is obtained through the pre-trained long short-term memory neural network, and product recommendations are made in combination with bank financial product data and historical purchase record data, so that data collection is easy and recommendation effect is good.

Description

Technical Field

[0001] The present invention relates to the financial field, and in particular to a method and device for recommending financial products based on interest rate prediction. Background Art

[0002] At present, the mainstream bank wealth management product recommendation algorithm is to recommend personalized wealth management products based on customer-based collaborative filtering algorithms. This algorithm must first improve the customer portrait, including the customer's basic information, financial status, risk preference and risk tolerance, investment experience, personal investment style, etc., and then calculate the basic characteristic mean of each bank's wealth management product corresponding to the customer group, compare the similarity of the customer group characteristic mean obtained based on the customer's historical purchase record and risk preference analysis, and finally select the wealth management products with the highest similarity as recommendations for this customer group.

[0003] The existing recommendation algorithms have two shortcomings. First, much of the data needed for customer profiling is inaccurate. For example, the customer's financial status is often lagging behind, and customers often do not seriously fill in relevant data on risk preferences, including changes in third-party escrow funds, credit limits, and other data used for risk assessment. Second, the interests of banks are not considered in conjunction with interest rate trends. For example, when expected interest rates rise, banks hope to recommend more long-term financial management products. Conversely, when expected interest rates fall, banks hope to recommend more short-term financial management products, which ultimately leads to unsatisfactory recommendation results. Summary of the invention

[0004] In view of the problems in the prior art, the present invention provides a method and device for recommending financial products based on interest rate prediction, an electronic device and a computer-readable storage medium, which can at least partially solve the problems in the prior art.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] In a first aspect, a method for recommending financial products based on interest rate prediction is provided, comprising:

[0007] Based on the financial product data and the corresponding historical purchase record data, obtain the support degree of two financial products being purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product also purchased another financial product, and the improvement degree of the application of the association rules of the two financial products;

[0008] Obtain bank interest rate forecast curve based on pre-trained long short-term memory neural network;

[0009] Calculate the recommendation degree of each group of financial products according to the financial product yield curve, the bank interest rate forecast curve and the improvement degree;

[0010] Financial products are recommended based on the support, the confidence and the recommendation degree.

[0011] Furthermore, obtaining the bank interest rate forecast curve includes:

[0012] Obtain interest rate influencing factors as features for predicting interest rates;

[0013] The interest rate of each month is classified into rising period, sideways period and falling period, and discretized using one-hot encoding as the feature for predicting interest rate;

[0014] All features are normalized and merged into a feature vector;

[0015] The feature vector is input as a prediction sample into a pre-trained long short-term memory neural network, and the output of the long short-term memory neural network is integrated into a bank interest rate prediction curve.

[0016] Furthermore, the interest rate influencing factors include: gross domestic product, consumer price index, producer price index, cash in circulation, narrow money, broad money, and RMB / USD exchange rate.

[0017] Furthermore, the interest rate influencing factors also include: GDP growth rate and narrow money year-on-year growth rate;

[0018] The factors affecting the acquisition rate include:

[0019] Use the feature construction and feature extraction techniques of industry feature engineering to explore the factors affecting the interest rate.

[0020] Furthermore, the method of obtaining, based on the financial product data and the corresponding historical purchase record data, the support degree of two financial products being purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product also purchased another financial product, and the improvement degree of application of the association rule of the two financial products, includes:

[0021] Preprocessing the financial product data and the corresponding historical purchase record data to obtain a customer purchase record table of financial products;

[0022] Calculate the support degree of two financial products being purchased by the same customer and the confidence degree of a customer who purchased one financial product also purchasing another financial product based on the record table of financial products purchased by the customer;

[0023] The degree of improvement of application of association rules of two financial products is calculated according to the record table of financial products purchased by the customer and the confidence level.

[0024] Furthermore, the calculation of the support degree of two financial products being purchased by the same customer and the confidence degree of a customer who purchased one financial product also purchased another financial product according to the record table of financial products purchased by the customer includes:

[0025] The support degree is obtained by dividing the number of customers who simultaneously purchased two financial products in the customer financial product purchase record table by the total number of customers;

[0026] The confidence level is obtained by dividing the number of customers who purchased two financial products at the same time in the customer financial product purchase record table by the number of customers who purchased one of the financial products.

[0027] Further, the calculating the improvement of the application of the association rules of the two financial products according to the record table of financial products purchased by the customer and the confidence level includes:

[0028] The degree of improvement is obtained by dividing the confidence by the proportion of customers who purchase another financial product in the record table of customers purchasing financial products.

[0029] In a second aspect, a financial product recommendation device based on interest rate prediction is provided, comprising:

[0030] The product analysis module obtains the support degree of two financial products purchased by the same customer in the customer's purchase records of financial products, the confidence degree of customers who purchased one financial product also purchased another financial product, and the improvement degree of the application of the association rules of the two financial products based on the financial product data and the corresponding historical purchase record data;

[0031] The bank interest rate forecast curve acquisition module obtains the bank interest rate forecast curve based on the pre-trained long short-term memory neural network;

[0032] A recommendation ranking module calculates the recommendation degree of each group of financial products according to the financial product yield curve, the bank interest rate forecast curve and the improvement degree;

[0033] A recommendation module recommends financial products according to the support, the confidence and the recommendation degree.

[0034] According to a third aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for recommending financial products based on interest rate forecasting when executing the program.

[0035] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method for recommending financial products based on interest rate prediction are implemented.

[0036] The method and device for recommending financial products based on interest rate prediction provided by the present invention are applicable to the financial field. The method comprises: obtaining, based on financial product data and corresponding historical purchase record data, the support degree of two financial products purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product and then purchased another financial product, and the improvement degree of the application of association rules of the two financial products; obtaining a bank interest rate prediction curve based on a pre-trained long short-term memory neural network; calculating the recommendation degree of each group of financial products based on the financial product yield curve, the bank interest rate prediction curve and the improvement degree; and recommending financial products based on the support degree, the confidence degree and the recommendation degree, wherein the bank interest rate prediction curve is obtained through the pre-trained long short-term memory neural network, and product recommendations are made in combination with bank financial product data and historical purchase record data, which not only makes data collection easy and of high quality, but also makes it possible to recommend products that are more in line with the interests of the bank in combination with interest rate trends; therefore, the technical solution can effectively avoid the risk of personalized recommendation failure caused by inaccurate customer information and personalized features, and can also select product recommendations that are beneficial to the bank in combination with interest rate predictions.

[0037] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0039] Figure 1 Schematic diagram of the architecture between the server S1 and the customer service device B1 in an embodiment of the present invention;

[0040] Figure 2 It is a schematic diagram of the architecture between the server S1, the customer service device B1 and the database server S2 in an embodiment of the present invention;

[0041] Figure 3 is a flow chart of a financial product recommendation method in an embodiment of the present invention;

[0042] Figure 4 The specific steps of step S100 in the embodiment of the present invention are shown;

[0043] Figure 5 An example of a record sheet of customers purchasing financial products is shown;

[0044] Figure 6 The specific steps of step S120 in the embodiment of the present invention are shown;

[0045] Figure 7 The specific steps of step S110 in the embodiment of the present invention are shown;

[0046] Figure 8 shows a product list in an embodiment of the present invention;

[0047] Fig. 9 The specific steps of step S300 in the embodiment of the present invention are shown;

[0048] Fig.10 An example of the purchase of all bank financial products within a specified period of time T in an embodiment of the present invention is shown;

[0049] Fig.11 The top ten examples of bank financial product portfolio recommendations that meet given support and confidence levels shown in the embodiments of the present invention are shown;

[0050] Fig.12 A relationship diagram of the top ten financial products ranked by the total number of times the financial products displayed in the embodiment of the present invention appear as the preceding item and the succeeding item - the main diagram;

[0051] Fig.13 The top ten wealth management products relationship diagram-node diagram displayed in the embodiment of the present invention are shown;

[0052] Fig.14 The top ten financial products relationship diagram and connection line diagram shown in the embodiment of the present invention are shown;

[0053] Fig.15 This is a structural diagram of the intelligent product combination recommendation system based on the correlation analysis of bank financial products sales volume of the present invention;

[0054] Fig.16 for Fig.15 Flow chart of processing steps of data analysis system 2;

[0055] Fig.17 for Fig.15 Module structure diagram of data analysis system 2;

[0056] Fig.18 is a structural block diagram of a financial product device in an embodiment of the present invention;

[0057] Fig.19 4 is a structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. 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 creative work should fall within the scope of protection of the present application.

[0059] 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.

[0060] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatuses.

[0061] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0062] The existing recommendation algorithms have two shortcomings. First, much of the data needed for customer profiling is inaccurate. For example, the customer's financial status is often lagging behind, and customers often do not seriously fill in relevant data on risk preferences, including changes in third-party escrow funds, credit limits, and other data used for risk assessment. Second, the interests of banks are not considered in conjunction with interest rate trends. For example, when expected interest rates rise, banks hope to recommend more long-term financial management products. Conversely, when expected interest rates fall, banks hope to recommend more short-term financial management products, which ultimately leads to unsatisfactory recommendation results.

[0063] In order to at least partially solve the above technical problems, the present invention provides a financial product recommendation method based on interest rate prediction. Product recommendations are made based on bank financial product data, historical purchase record data, and the bank's interest rate prediction curve. Not only is data collection easy and the data quality high, but the interest rate trend can also be combined to recommend products that are more in line with the bank's interests. This can effectively avoid the risk of personalized recommendation failures due to inaccurate customer information and personalized characteristics, and can also select product recommendations that are beneficial to the bank in combination with interest rate predictions.

[0064] In view of this, the present application provides a financial product recommendation device based on interest rate prediction, which can be a server S1, see Figure 1 , the server S1 can be connected to at least one customer service device B1 in communication. The server S1 obtains the support degree of two financial products purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product and then purchased another financial product, and the improvement degree of the application of the association rules of the two financial products based on the financial product data and the corresponding historical purchase record data; obtains the bank interest rate prediction curve based on the pre-trained long short-term memory neural network; calculates the recommendation degree of each group of financial products based on the financial product yield curve, the bank interest rate prediction curve and the improvement degree; and recommends financial products based on the support degree, the confidence degree and the recommendation degree. Then, the server S1 can send the recommendation result online to the customer service device B1. The customer service device B1 can receive the recommendation result online.

[0065] Also, see Figure 2 The server S1 may also be connected to at least one database server S2 for communication, and the database server S2 is used to store the financial product data and the corresponding historical purchase record data. The database server S2 sends the financial product data and the corresponding historical purchase record data to the server S1 online, and the server S1 may receive the financial product data and the corresponding historical purchase record data online.

[0066] It is understandable that the customer service device B1 may include a smart phone, a tablet electronic device, a portable computer, a desktop computer, a personal digital assistant (PDA), etc.

[0067] The server and the client device may communicate using any suitable network protocol, including network protocols that have not yet been developed on the date of filing this application. The network protocol may include, for example, TCP / IP protocol, UDP / IP protocol, HTTP protocol, HTTPS protocol, etc. Of course, the network protocol may also include, for example, RPC protocol (Remote Procedure Call Protocol) and REST protocol (Representational State Transfer) used on top of the above protocols.

[0068] Figure 3 is a flow chart of a method for recommending financial products based on interest rate prediction in an embodiment of the present invention; Figure 3 As shown, the financial product recommendation method based on interest rate prediction may include the following contents:

[0069] Step S100: According to the financial product data and the corresponding historical purchase record data, obtain the support degree of two financial products purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a customer who purchased one financial product also purchased another financial product, and the improvement degree of the application of the association rule of the two financial products;

[0070] Specifically, the wealth management product data includes the names of the bank's multiple wealth management products and their corresponding product numbers, returns, purchase thresholds, durations, investment types and other parameters; the historical purchase record data includes the customer number, customer information and the corresponding names of the wealth management products purchased and their corresponding product numbers, purchase time, purchase amount and other parameters.

[0071] Step S200: obtaining a bank interest rate prediction curve based on a pre-trained long short-term memory neural network;

[0072] Step S300: Calculating the recommendation degree of each group of financial products according to the financial product yield curve, the bank interest rate forecast curve and the improvement degree;

[0073] Among them, the bank interest rate forecast curve is the interest rate trend obtained by the relevant departments of the bank based on a comprehensive analysis of market changes.

[0074] Step S400: recommending financial products according to the support, the confidence and the recommendation degree.

[0075] Specifically, the recommended financial products can display the recommended financial product information on the one hand, and on the other hand, can also display the related financial product information, so as to provide sufficient guidance for customer service personnel and help customer service personnel recommend to customers.

[0076] It is worth noting that, compared with the various data required for customer profiling, various information on wealth management products, as data automatically generated by banks when processing business, is easier to be collected by the bank itself and has the highest data quality. Banks have a certain amount of experience in predicting interest rate trends and can recommend products that are more in line with the bank's interests based on expected interest rate trends. Therefore, based on the sales volume of bank wealth management products, the combination rules between bank wealth management products are explored. While looking for the correlation between products, product recommendations are weighted based on interest rate expectations. This can effectively avoid the risk of personalized recommendation failures caused by inaccurate customer information and personalized characteristics, and can also select product recommendations that are beneficial to the bank based on interest rate forecasts, thereby achieving intelligent recommendations based on the correlation between bank wealth management product sales and weighted by expected interest rates.

[0077] In an optional embodiment, step S200 may include the following contents:

[0078] (1) Obtain interest rate influencing factors as features for predicting interest rates;

[0079] Specifically, the data of interest rate influencing factors in the past thirty years, including but not limited to gross domestic product (GDP), consumer price index (CPI), producer price index (PPI), cash in circulation (M0), narrow money (M1), broad money (M2), RMB / USD exchange rate, etc., are arranged in chronological order;

[0080] In addition, we use the feature construction and feature extraction methods of feature engineering in the industry to further explore the factors affecting interest rates, such as GDP growth rate, M1 year-on-year growth rate, etc., and use these factors as features for predicting interest rates;

[0081] (2) Classify the interest rate of each month into rising period, sideways period, and falling period, and use one-hot encoding to discretize them as features for predicting interest rates;

[0082] (3) All features are normalized and merged into a feature vector;

[0083] It is worth noting that all features are standardized, and the standardized formula is:

[0084]

[0085] Among them, V x V is the specific value of GDP, CPI, PPI, M0, M1, M2, RMB / USD exchange rate, etc. in a certain month. max V is the maximum value of GDP, CPI, PPI, M0, M1, M2, RMB / USD exchange rate, etc. in the past 20 years. min is the minimum value, V n It is the numerical value after the specific value is standardized.

[0086] Example: The maximum value of CPI in the past 20 years V max 12%, minimum V min is 2%. If the CPI value of a certain month is V x is 5%, then the standardized value

[0087] For example, all the standardized features are combined into a feature vector X(V GDP ,V CPI ,V PPI ,V M0 ,V M1 ,V M2 ,V er ……).

[0088] (4) Perform window segmentation on the features, that is, use a time step of n to determine the use of data from n months before a certain month to predict the interest rate of the next month;

[0089] (5) The feature vector X(V GDP ,V CPI ,V PPI ,V M0 ,V M1 ,V M2 ,V er ……) is used as the prediction sample input to the input layer of the pre-trained long short-term memory neural network LSTM, and the output layer outputs the predicted interest rate result for the next month. The output of the long short-term memory neural network is integrated into the bank interest rate prediction curve R(T x ).

[0090] Figure 4 Detailed steps of step S100 in the embodiment of the present invention are shown; Figure 4 As shown, step S100 may include the following contents:

[0091] Step S110: pre-processing the financial product data and the corresponding historical purchase record data to obtain a customer purchase record table of financial products;

[0092] Within a specified period of time T, the customer's purchase records are integrated into a list L. Each record in the list L contains a customer number and a financial product, such as Figure 5 As shown; if customer a purchased 5 financial products within time T, then the list will have 5 records about customer a.

[0093] Step S120: Calculate the support degree of two financial products being purchased by the same customer and the confidence degree of a user who purchased one financial product also purchased another financial product according to the record table of financial products purchased by the customer;

[0094] In an alternative embodiment, see Figure 6 , the step S120 may include the following contents:

[0095] Step S121: the support degree is obtained by dividing the number of customers who simultaneously purchased two financial products in the customer financial product purchase record table by the total number of customers.

[0096] Specifically, the calculation formula for the support of financial product A and financial product B purchased by the same customer is as follows:

[0097] R support =Number of customers who purchase financial products A and B at the same time / Total number of customers=P(AB) / P(ALL).

[0098] Step S122: The confidence level is obtained by dividing the number of customers who purchased both financial products in the customer financial product purchase record table by the number of customers who purchased one of the financial products.

[0099] Specifically, the confidence level is the proportion of people who purchased financial product A who purchased financial product B. The calculation formula is as follows:

[0100] R confidence =The number of customers who purchase financial products A and B at the same time / the total number of customers who purchase financial product A=P(AB) / P(A).

[0101] Step S130: Calculate the degree of improvement of the application of the association rules of two financial products according to the record table of financial products purchased by the customer and the confidence level.

[0102] Specifically, the degree of improvement is obtained by dividing the confidence by the proportion of customers who purchase another financial product in the record table of customers purchasing financial products.

[0103] Among them, the improvement degree of application of association rules of financial products A and B is calculated, that is, the ratio of the results of applying association rules to those of not applying them. The calculation formula is as follows:

[0104] R lift =The proportion of people who purchased financial product A who purchased financial product B / the proportion of customers who purchased financial product B = P(B|A) / P(B|ALL).

[0105] In an alternative embodiment, see Figure 7 , the step S110 may include the following contents:

[0106] Step S111: merging multiple financial products according to the attribute parameters of each financial product to obtain a product list;

[0107] Specifically, within a specified period of time T, the wealth management products with the same or very similar returns, purchase thresholds, duration, investment types, etc. are merged to obtain a product list. The basis is that the bank's wealth management products have a sales period, and they are basically no longer sold after the fundraising is completed, resulting in many products that are only different in name or number of periods, but are actually the same wealth management product; merging these products is conducive to improving the stability of data and algorithms; after merging, a product list is obtained, such as Figure 8 shown.

[0108] Step S112: obtaining a record table of the customer's purchase of financial products based on the product list and the historical purchase record data.

[0109] Specifically, during integration, the products purchased by the customer in the historical purchase record data are first integrated according to the product list, and then a customer number and a corresponding financial product purchased are stored as a record in the customer purchase product record table.

[0110] In an alternative embodiment, see Fig. 9 , the step S300 may include the following contents:

[0111] Step S310: Calculating the recommendation weight of each group of financial products according to the financial product yield curve and the bank interest rate forecast curve;

[0112] Specifically, combined with the bank's forecast of interest rates; assuming that the term of financial product X is T X , term T X The domestic bank interest rate forecast curve is R(T X ), the yield curve of financial product X over time is G(T X ), then the recommendation weight W(X) of financial product X is calculated as follows:

[0113]

[0114] If the income curve of financial product X changes over time is G(T X ) does not exist, then use the estimated curve G'(T X ) instead.

[0115] Step S320: Calculate the recommendation degree of each group of financial products according to the recommendation degree weight and the improvement degree.

[0116] Specifically, according to the lifting degree R lift 、Recommendation weight W(X) calculates recommendation R recommend , the calculation formula is as follows:

[0117] R recommend =R lift ×W(X).

[0118] In an optional embodiment, step S400 may include:

[0119] Step I: After eliminating the financial product combinations whose support is lower than the preset minimum support threshold and whose confidence is lower than the minimum confidence threshold, the remaining financial product combinations are sorted according to the recommendation degree and then displayed.

[0120] Step II: Count the number of occurrences of each financial product in the sorted financial product portfolio;

[0121] Step III: Sort the financial products according to their occurrence frequency;

[0122] Step IV: Display the relevant data of the first preset number of financial products among the financial products sorted by the number of occurrences through a relationship diagram.

[0123] For example, according to the support R calculated above support, confidence R confidence , Recommended R recommend , pre-set the minimum support threshold, such as minS = 0.01, and pre-set the minimum confidence threshold, such as minC = 0.05, and eliminate the financial product combinations that are lower than the minimum support threshold and the minimum confidence threshold. Sort the financial product combinations that meet the conditions by recommendation. The financial product combination contains a front item 1 (i.e., the financial product at the front of the combination) and a back item 2 (i.e., the financial product at the back of the combination). Count the total number of times each financial product appears as item 1 and item 2, and sort the financial products from large to small according to the total number of times to get the top ten financial products.

[0124] After statistical processing, the analysis results can be displayed to customer service personnel, such as displaying the purchase status of all bank financial products within a specified period of time T, such as Fig.10 shown.

[0125] Of course, within a specified period of time T, the bank wealth management product portfolio that meets the given support and confidence will be sorted from large to small according to the recommendation degree, and the top ten will be displayed. The displayed content includes a preceding item represented by item1, a subsequent item represented by item2, the number of people who purchase item2 under the premise of purchasing item1, support, confidence, and recommendation, such as Fig.11 shown.

[0126] It is worth noting that the top ten bank financial products can be displayed through a relationship diagram, such as Fig.12 As shown; in the relationship diagram, each node represents a bank's wealth management product, and the size of the node is proportional to the total number of times the corresponding wealth management product is the antecedent and the consequent; the edge between any two nodes represents the existence of a correlation between the two wealth management products, that is, the sales of one of the wealth management products will affect the sales of the other wealth management product.

[0127] In addition, in the relationship diagram, move the mouse to the node to be viewed, and the name and sales volume of the financial product represented by the node will be highlighted; at the same time, all other financial products related to this financial product will be highlighted, such as Fig.13 As shown; move the mouse to the edge to be viewed, highlight the two nodes connected by the edge, and use the floating text "XXX>xxx:n" to indicate that among the customers who purchased financial product XXX, the number of customers who purchased financial product XXX is n, such as Fig.14 shown.

[0128] When a customer purchases a financial product listed in the preceding paragraph, Fig.10Based on the displayed content, the corresponding subsequent financial products are searched for as recommendations; several financial products with many association rules can be selected as a combination and recommended to customers at the same time.

[0129] Fig.15 This is a structural diagram of the intelligent product combination recommendation system based on the correlation analysis of bank financial products sales in the present invention; the intelligent product combination recommendation system is used to implement the financial product recommendation method provided in the embodiment of the present invention, and the intelligent product combination recommendation system includes: data collection system 1, data analysis system 2, and data display system 3, the data collection system 1 is connected to the data analysis system 2, and the data analysis system 2 is connected to the data display system 3. The data collection system 1 is deployed in each branch and outlet of the bank, the data analysis system 2 is deployed in the bank's technology department, the data display system 3 is deployed in the bank's technology department, and the front-end page of the data display system 3 is accessible to all levels of bank institutions (head office, first-level branch, second-level branch, branch, outlet).

[0130] Specifically, the data collection system 1 is deployed in each branch and outlet of the bank, and is responsible for collecting records of all customers purchasing any one or more financial products. The data analysis system 2 is deployed in the bank's technology department, and is responsible for preprocessing, calculating, and analyzing the bank's financial product purchase record data collected by the data collection system 1 to obtain results.

[0131] Fig.17 2 is a module structure diagram of the data analysis system 2 in the system of the present invention. The data analysis system 2 includes a data preprocessing module 21, a data calculation module 22, and a data statistics module 23, wherein the data preprocessing module 21 is connected to the data calculation module 22, and the data calculation module 22 is connected to the data statistics module 23.

[0132] Specific:

[0133] For the processing flow of the data preprocessing module 21, see Fig.16 ,as follows:

[0134] Step S201: Merge financial products with the same or similar returns, purchase thresholds, durations, investment types, etc. within a specified period to obtain a product list;

[0135] Specifically, the wealth management products with the same or very similar returns, purchase thresholds, duration, investment types, etc. within a specified period of time T are merged to obtain a product list. The reason is that the bank's wealth management products have a sales period, and they are basically no longer sold after the fundraising is completed, resulting in many products that are only different in name or number of periods, but are actually the same wealth management product; merging these products is conducive to improving the stability of data and algorithms; after merging, a product list is obtained.

[0136] Step S202: The customer's purchase records are integrated into a list L, where each record in the list L contains a customer number and a financial product;

[0137] Specifically, within a specified period of time T, the customer's purchase records are integrated into a list L, and each record in list L contains a customer number and a financial product; if customer a purchases 5 financial products within time T, then the list has 5 records about customer a.

[0138] The processing flow of the data calculation module 22 is as follows:

[0139] Step S203: Calculate support;

[0140] Specifically, according to the list L obtained in step S202, the support degree of the financial products being purchased in pairs is calculated; the calculation formula for the support degree of financial products A and B being purchased by the same customer is as follows:

[0141] R support = Number of customers who purchase financial products A and B at the same time / Total number of customers = P(AB) / P(ALL)

[0142] Step S204: calculating the confidence level;

[0143] Specifically, according to the list L obtained in step S202, the confidence of purchasing financial product A and financial product B at the same time is calculated, that is, the proportion of people who purchased financial product A who purchased financial product B is calculated as follows:

[0144] R confidence = Number of customers who purchase both financial products A and B / Total number of customers who purchase financial product A = P(AB) / P(A)

[0145] Step S205: Calculate the lifting degree;

[0146] According to the list L obtained in step S202 and the calculation results obtained in steps S203 and S204, the improvement degree of application of the association rules of financial products A and B is calculated, that is, the ratio of the results generated by applying the association rules to those generated by not applying the association rules. The calculation formula is as follows:

[0147] R lift = The proportion of people who purchased financial product A who purchased financial product B / The proportion of customers who purchased financial product B = P(B|A) / P(B|ALL)

[0148] Step S206: Calculate the recommendation weight based on the bank's prediction of interest rates;

[0149] Specifically, combined with the bank's forecast of interest rates; assuming that the term of financial product X is T X , term TX The domestic bank interest rate forecast curve is R(T X ), the yield curve of financial product X over time is G(T X ), then the recommendation weight W(X) of financial product X is calculated as follows:

[0150]

[0151] If the income curve of financial product X changes over time is G(T X ) does not exist, then use the estimated curve G'(T X ) instead.

[0152] Step S207: Calculate the recommendation degree according to the promotion degree and the recommendation degree weight;

[0153] Specifically, according to the lifting degree R in step S205 lift , the recommendation weight W(X) of step S206, calculate the recommendation R recommend , the calculation formula is as follows:

[0154] R recommend =R lift ×W(X)

[0155] The processing flow of the data statistics module 23 is as follows:

[0156] Step S208: Eliminate financial products that do not meet the given support and confidence levels, and sort the financial product combinations that meet the conditions according to the recommendation level;

[0157] Specifically, the support R calculated in step S203, step S204, and step S207 is support , confidence R confidence , Recommended R recommend , set the minimum support threshold minS = 0.01, the minimum confidence threshold minC = 0.05, eliminate the financial product combinations with support and confidence levels lower than the set values, and sort the financial product combinations that meet the conditions according to the recommendation level. The financial product combination contains a preceding item item1 and a succeeding item item2.

[0158] Step S209: Count the total number of times each financial product appears as the preceding term or the succeeding term, and sort the financial products from largest to smallest according to the total number of times to obtain the top ten financial products.

[0159] Specifically, according to the calculation result of step S208, the total number of times each financial product appears as item1 and item2 is counted, and the financial products are sorted from large to small according to the total number of times to obtain the top ten financial products.

[0160] The backend of data display system 3 is deployed in the bank's technology department, and the front-end page is accessible to all levels of bank institutions (head office, first-level branches, second-level branches, sub-branches, and outlets); data display system 3 is responsible for displaying the processing results of data analysis system 2.

[0161] For example, the purchase status of all bank financial products within a specified period of time T is displayed, or the bank financial product portfolio that meets a given support and confidence level within a specified period of time T is sorted from high to low according to the recommendation level, and the top ten are displayed. The displayed content includes a preceding item represented by item1, a subsequent item represented by item2, the number of people who purchased item2 under the premise of purchasing item1, the support level, the confidence level, and the recommendation level.

[0162] In addition, the data display system 3 can also display the top ten bank wealth management products through a relationship diagram based on the calculation result of step S207; in the relationship diagram, each node represents a wealth management product of a bank, and the size of the node is proportional to the total number of times the corresponding wealth management product is the antecedent and the subsequent item; the edge between any two nodes represents that there is a correlation between the two wealth management products, that is, the sales of one of the wealth management products will affect the sales of the other wealth management product.

[0163] In the relationship diagram, move the mouse to the node to be viewed, and the name and sales volume of the wealth management product represented by the node will be highlighted; at the same time, all other wealth management products associated with this wealth management product will be highlighted. Move the mouse to the edge to be viewed, and the two nodes connected by the edge will be highlighted. Use the floating text "XXX>xxx:n" to indicate that among the customers who purchased wealth management product XXX, the number of customers who purchased wealth management product xxx is n.

[0164] In summary, the wealth management product recommendation method provided in the embodiment of the present invention directly analyzes the sales data of bank wealth management products, parses the associations generated by customers purchasing bank wealth management products, and mines the combination rules between various products; then, the interest rate prediction curve and product returns are combined to calculate the recommendation degree; it has the advantages of accurate data source, convenient data acquisition, fast analysis, intuitive result display and multiple application methods of analysis results, and can effectively avoid the risk of personalized recommendation failure caused by inaccurate customer information and personalized characteristics. At the same time, it combines interest rate prediction to select product recommendations that are beneficial to the bank, providing a strong reference for bank account managers to recommend wealth management products.

[0165] Based on the same inventive concept, the embodiments of the present application also provide a financial product recommendation device, which can be used to implement the method described in the above embodiments, as described in the following embodiments. Since the principle of solving the problem by the financial product recommendation device is similar to that of the above method, the implementation of the financial product recommendation device can refer to the implementation of the above method, and the repeated parts will not be repeated. As used below, the term "unit" or "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware, may also be conceived and implemented.

[0166] Fig.18 is a structural block diagram of the financial product recommendation device in an embodiment of the present invention. Fig.18 As shown, the financial product recommendation device specifically includes: a product analysis module 10, a bank interest rate prediction curve acquisition module 20, a recommendation degree ranking module 30 and a recommendation module 40.

[0167] The product analysis module 10 obtains the support degree of two financial products purchased by the same customer in the customer's purchase record of financial products, the confidence degree of a user who purchased one financial product also purchased another financial product, and the promotion degree of application of the association rule of the two financial products based on the financial product data and the corresponding historical purchase record data;

[0168] The bank interest rate prediction curve acquisition module 20 acquires the bank interest rate prediction curve based on the pre-trained long short-term memory neural network;

[0169] The recommendation ranking module 30 calculates the recommendation degree of each group of financial products according to the financial product yield curve, the bank interest rate prediction curve and the improvement degree;

[0170] The recommendation module 40 recommends financial products according to the support, the confidence and the recommendation degree.

[0171] Compared with the various data required for customer profiling, various information on wealth management products, as data automatically generated by banks when processing business, is easier to be collected by the bank itself and has the highest data quality. Banks have a certain amount of experience in predicting interest rate trends and can recommend products that are more in line with the bank's interests based on expected interest rate trends. Therefore, based on the sales volume of bank wealth management products, the combination rules between bank wealth management products are explored. While looking for the correlation between products, product recommendations are weighted based on interest rate expectations. This can effectively avoid the risk of personalized recommendation failures caused by inaccurate customer information and personalized characteristics, and can also select product recommendations that are beneficial to the bank based on interest rate forecasts, thereby achieving intelligent recommendations based on the correlation between bank wealth management product sales and weighted by expected interest rates.

[0172] The devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is an electronic device, and the electronic device may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0173] In a typical example, the electronic device specifically includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the above-mentioned financial product recommendation method are implemented.

[0174] Reference below Fig.19 , which shows a schematic structural diagram of an electronic device 600 suitable for implementing an embodiment of the present application.

[0175] like Fig.19 As shown, electronic device 600 includes a central processing unit (CPU) 601, which can perform various appropriate operations and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage portion 608 into a random access memory (RAM) 603. In RAM 603, various programs and data required for the operation of system 600 are also stored. CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to bus 604.

[0176] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed as needed as the storage section 608.

[0177] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above-mentioned financial product recommendation method when executed by a processor.

[0178] In such an embodiment, the computer program may be downloaded and installed from a network via the communication section 609 , and / or installed from the removable medium 611 .

[0179] 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.

[0180] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0181] 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.

[0182] 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.

[0183] 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 The steps for the functions specified in one or more boxes.

[0184] 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.

[0185] 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.

[0186] The present application may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0187] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0188] 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 recommending financial products based on interest rate prediction, characterized in that: include: Based on the wealth management product data and the corresponding historical purchase records, obtain the support level of two wealth management products being purchased by the same customer, the confidence level of a customer who purchased one wealth management product also purchasing another wealth management product, and the improvement level of the application of the association rules between the two wealth management products; Obtain bank interest rate prediction curve based on pre-trained long short-term memory neural network; Calculating the recommendation degree of each group of financial products based on the financial product yield curve, the bank interest rate forecast curve and the improvement degree; Recommending financial products based on the support, the confidence, and the recommendation; The method of obtaining a bank interest rate prediction curve based on a pre-trained long short-term memory neural network includes: Obtain interest rate influencing factors as features for predicting interest rates; The monthly interest rate is classified into rising period, sideways period and falling period, and discretized using one-hot encoding as the feature for predicting interest rate; All features are normalized and merged into a feature vector; The feature vector is input into a pre-trained long short-term memory neural network as a prediction sample, and the output of the long short-term memory neural network is integrated into a bank interest rate prediction curve.

2. The method for recommending financial products based on interest rate prediction according to claim 1, characterized in that: The interest rate influencing factors include: gross domestic product, consumer price index, producer price index, cash in circulation, narrow money, broad money, and the RMB / USD exchange rate.

3. The method for recommending financial products based on interest rate prediction according to claim 2, characterized in that: The interest rate influencing factors also include: GDP growth rate and year-on-year growth rate of narrow money; The factors affecting the acquisition interest rate include: Use the feature construction and feature extraction technology of industry feature engineering to explore the factors affecting the interest rate.

4. The method for recommending financial products based on interest rate prediction according to claim 1, characterized in that: The method of obtaining, based on the wealth management product data and the corresponding historical purchase record data, the support level of two wealth management products being purchased by the same customer, the confidence level of a customer who purchased one wealth management product also purchasing another wealth management product, and the improvement level of applying the association rules between the two wealth management products in the wealth management product purchase records of the customer, includes: Preprocessing the financial product data and the corresponding historical purchase record data to obtain a customer purchase record table of financial products; Calculate, based on the customer purchase record table of financial products, the support level of two financial products being purchased by the same customer and the confidence level of a customer who purchased one financial product also purchasing another financial product; The improvement degree of application of the association rules of the two financial products is calculated based on the customer purchase record table of financial products and the confidence level.

5. The method for recommending financial products based on interest rate prediction according to claim 4, characterized in that: The calculating, based on the record table of financial products purchased by the customer, the support degree of two financial products being purchased by the same customer and the confidence degree of a customer who purchased one financial product also purchasing another financial product includes: The support degree is obtained by dividing the number of customers who purchased two financial products at the same time in the customer financial product purchase record table by the total number of customers; The confidence level is obtained by dividing the number of customers who purchased both financial products in the customer financial product purchase record table by the number of customers who purchased one of the financial products.

6. The method for recommending financial products based on interest rate prediction according to claim 5, characterized in that: The calculating the improvement of the application of the association rules of the two financial products according to the record table of financial products purchased by the customer and the confidence level includes: The confidence level is divided by the proportion of customers who purchased another financial product in the customer financial product purchase record table to obtain the lift.

7. A financial product recommendation device based on interest rate prediction, characterized in that: include: The product analysis module uses financial product data and corresponding historical purchase records to obtain information about the support level of two financial products purchased by the same customer, the confidence level of customers who purchased one financial product also purchasing another, and the improvement level of the application of association rules between the two financial products. The bank interest rate forecast curve acquisition module obtains the bank interest rate forecast curve based on the pre-trained long short-term memory neural network; A recommendation ranking module calculates the recommendation degree of each group of financial products based on the financial product yield curve, the bank interest rate forecast curve and the improvement degree; A recommendation module, which recommends financial products based on the support, the confidence, and the recommendation degree; The method of obtaining a bank interest rate prediction curve based on a pre-trained long short-term memory neural network includes: Obtain interest rate influencing factors as features for predicting interest rates; The monthly interest rate is classified into rising period, sideways period and falling period, and discretized using one-hot encoding as the feature for predicting interest rate; All features are normalized and merged into a feature vector; The feature vector is input into a pre-trained long short-term memory neural network as a prediction sample, and the output of the long short-term memory neural network is integrated into a bank interest rate prediction curve.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method for recommending financial products based on interest rate prediction according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for recommending financial products based on interest rate prediction according to any one of claims 1 to 6 are implemented.

Images