User portrait construction method based on financial transaction data

Abstract

The invention discloses a user portrait construction method based on financial transaction data, and relates to the technical field of finance, and the method comprises the steps: training the number of historical financial transactions of a user, obtaining the personality type of the user in the aspect of investment based on the historical financial transaction data of the user, obtaining the user portrait of the user, and obtaining the user portrait of the user according to the obtained user portrait. Therefore, the personality types owned by the users in the aspect of investment are subjected to weight division, which user type the users belong to is judged, and a proper investment scheme is pushed to the users according to the weight division of different personality types; and meanwhile, multi-dimensional analysis can be performed on the user on the basis of historical financial transaction data of the user by utilizing the tree model, so that the finally obtained user portrait is more appropriate to the actual condition of the user, and an investment scheme provided for the user is more consistent with the demand of the user.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and specifically to a method for constructing a user portrait based on financial transaction data. Background Art

[0002] With the rapid development of Internet technology, the amount of payment-related data on Internet financial platforms is growing exponentially, and correspondingly, various suppliers providing Internet financial services are emerging like mushrooms after rain. Thus, the data gradually accumulated in the Internet financial network will be huge now and in the near future. Therefore, how to mine and analyze these massive financial payment data to improve the quality of financial services provided by suppliers to users is particularly important; How to obtain the investor type of a user based on the user's financial transaction data and then recommend a suitable investment plan for the user according to the investor type of the user is a problem that we need to solve. For this reason, a method for constructing a user portrait based on financial transaction data is provided herein. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for constructing a user portrait based on financial transaction data.

[0004] The purpose of the present invention can be achieved by the following technical solutions: A method for constructing a user portrait based on financial transaction data, comprising the following steps: Step 1: Establish a user transaction information database, and record the user's historical financial transaction information through the user transaction information database; Step 2: Preprocess the obtained user's historical financial transaction information to obtain financial sample data associated with the user; Step 3: Construct a user portrait for the user according to the obtained financial sample data.

[0005] Further, the process of establishing the user transaction information database includes: Set up a registration unit and a login unit in the financial trading platform for the user to conduct identity registration verification for financial transactions, so that the user can register identity information to obtain the permission to enter the financial trading platform; After obtaining the permission to enter the financial trading platform, a user transaction information database associated with the registered account is established according to the registered account, and the financial transaction information of the user in the financial trading platform is recorded through the user transaction information database.

[0006] Further, the process of the user registering identity information includes: Input the user's personal basic information through the registration unit, and the financial trading platform audits the input personal basic information and outputs an audit result; After the review is passed, a login account and a login password are generated based on the mobile phone number in the input personal basic information; By inputting the login account and the login password into the login unit, the user can obtain the permission to enter the financial trading platform.

[0007] Furthermore, the process of preprocessing the obtained user historical financial transaction information includes: Read each transaction record and the transaction amount of each transaction record in the user transaction information database; Obtain the number of overdraft transactions and the overdraft limit of each overdraft transaction, and associate each overdraft limit with the transaction record respectively; Obtain the transaction type corresponding to each transaction record, classify the transaction records according to the transaction type, and obtain a multi-dimensional transaction record data set according to its transaction amount; Perform normalization processing on the multi-dimensional transaction record data set so that each column of data in the multi-dimensional transaction record data set is between [0, 1], and record the normalized data set as financial sample data.

[0008] Furthermore, in the process of obtaining transaction records and overdraft transactions, a behavior time window for obtaining transaction records and overdraft transactions is set. The time of the behavior time window is T, and the time T of the behavior time window is the duration between the t1 moment and the t2 moment, where the t1 moment is the current moment and the t2 moment is the past moment with a time interval of T from the t1 moment.

[0009] Furthermore, the process of creating a user profile for the user includes: Build a tree model, and use the financial sample data as input and input it into the tree model to obtain a feature set Obtain a residual data set according to the difference between the obtained feature set and the financial sample data; Use the obtained residual data set as the training set and continuously train it until the obtained residual is 0 or reaches the upper limit of the training times of the tree model, obtain the finally extracted feature set, mark the finally extracted feature set as the financial feature set, and obtain the user profile according to the obtained financial feature set.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: By training the historical financial transactions of users, the personality types of users in terms of investment based on the historical financial transaction data of users are obtained, and user portraits of users are obtained. According to the obtained user portraits, the weight division of the personality types of users in terms of investment is carried out, and it is judged what type of user the user belongs to. According to the weight division of different personality types, a suitable investment plan is promoted for the user; at the same time, using the tree model, multi-dimensional analysis of the user can be carried out based on the historical financial transaction data of the user, so that the finally obtained user portrait is more in line with the actual situation of the user, and further the investment plan provided for the user is more in line with the needs of the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 is the schematic diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0012] As Figure 1 shown, a method for constructing a user portrait based on financial transaction data includes the following steps: Step 1: Establish a user transaction information database, and record the historical financial transaction information of users through the user transaction information database; It should be further noted that, in the specific implementation process, the establishment process of the user transaction information database includes: Set a registration unit and a login unit in the financial trading platform for users to perform identity registration verification for financial transactions, so that users can register their identity information to obtain the permission to enter the financial trading platform; Specifically, input the personal basic information of the user through the registration unit, and the financial trading platform audits the input personal basic information and outputs the audit result; it should be further noted that, in the specific implementation process, the personal basic information includes name, gender, age, ID card information, the user's own bank card for transactions, and the mobile phone number for real-name authentication; After the audit is passed, a login account and a login password are generated according to the mobile phone number in the input personal basic information, and the generated login account and login password are sent to the corresponding mobile terminal; By inputting the obtained login account and login password into the login unit, the user can obtain the permission to enter the financial trading platform; After obtaining the permission to enter the financial trading platform, a user transaction information database associated with the registration account is established according to the registration account, and the financial transaction information of the user in the financial trading platform is recorded through the user transaction information database. The financial transaction information includes transaction amount, payment record, overdraft record, transaction voucher, securities entrustment, transaction, position information, insurance policy information, claim information, etc.

[0013] Step 2: Preprocess the obtained historical financial transaction information of the user to obtain financial sample data associated with the user; It should be further noted that, in the specific implementation process, the process of preprocessing the obtained historical financial transaction information of the user includes: Read each transaction record in the user transaction information database and label it as i, where i = 1, 2,..., n, and n is an integer; Obtain the transaction amount of each transaction record, and denote the transaction amount corresponding to the transaction record with label i as JE i ; Obtain the number of overdraft transactions among them, and label each overdraft transaction as k, where k = 1, 2,..., m, and m is an integer, and m ≤ n; It should be further noted that, in the specific implementation process, in the process of obtaining transaction records and overdraft transactions, a behavior time window for obtaining transaction records and overdraft transactions is set, and the time of the behavior time window is T. The time T of the behavior time window is the duration between the t1 moment and the t2 moment, where the t1 moment is the current moment, and the t2 moment is the past moment with a time interval of T from the t1 moment; Obtain the overdraft amount of each overdraft transaction, and denote the overdraft amount with label k as TE k , and associate each overdraft amount with the transaction record respectively; It should be further noted that, in the specific implementation process, each overdraft amount is associated with at least one transaction record. When there are consecutive transactions that cause the overdraft amount to increase, then associate the overdraft amount corresponding to the corresponding overdraft transaction number with the consecutive transaction numbers; Embodiment

[0014] Suppose there are 3 transaction records a1, a2, a3, and the transaction amounts corresponding to each transaction record are a1 = 10, a2 = 20, and a3 = 30 respectively; There are 2 overdraft transactions b1, b2, and the overdraft amounts of each overdraft transaction are b1 = 10 and b2 = 20 in sequence; Among them, b1 is associated with a2, that is, after completing the a2nd transaction, there is an overdraft amount b1; b2 is associated with a2 and a3, that is, the overdraft amount of b2 is formed by the two transactions a2 and a3, and so on; that is, when there are consecutive transactions that cause the overdraft amount to increase, then associate the overdraft amount corresponding to the corresponding overdraft transaction number with the consecutive transaction numbers.

[0015] Obtain the transaction type corresponding to each transaction record, classify the transaction records according to the transaction type, and obtain a multi-dimensional transaction record dataset X according to its transaction amount, where X = , where l represents the l-th dimension, yp represents the yp-th transaction record within the same dimension, and y1 + y2 + …… + yp = n; Normalize the multi-dimensional transaction record dataset so that each column of data in the multi-dimensional transaction record dataset is between [0, 1], thereby obtaining the normalized dataset. Denote the normalized dataset as financial sample data, and denote the obtained financial sample data as M, where M = ; Step 3: Create a user profile for the user based on the obtained financial sample data; Specifically, establish a tree model, and use the financial sample data as input and input it into the tree model to obtain a feature set. Denote the obtained feature set as C; where C = f(M, W) = ; where M is the input sample, that is, the financial sample data, z is the classification and regression tree, W is the parameter of the classification and regression tree, and a is the weight of each classification and regression tree; Based on the difference between the obtained feature set C and the financial sample data, obtain a residual dataset. Denote the residual dataset as D; It should be further noted that in the specific implementation process, the calculation formula for the residual dataset is E = ; Use the obtained residual dataset as the training set and continuously train until the obtained residual is 0 or reaches the upper limit of the training times of the tree model, and obtain the finally extracted feature set. Mark the finally extracted feature set as the financial feature set; Obtain the user profile based on the obtained financial feature set, judge the user type according to the user profile, and recommend a suitable investment plan for the user according to the user type; It should be further noted that in the specific implementation process, user types include conservative investors, prudent investors, balanced investors, and aggressive investors, etc.; It should be further noted that in the specific implementation process, the user type that the same user has can be multiple. Through the obtained user profile, obtain the proportion weights of the user in different user types, so as to generate a mixed investment plan for the user; Embodiment

[0016] Suppose a user's user types include conservative investors and aggressive investors, where the weight of conservative investors is 80% and the weight of aggressive investors is 20%; Then there are at least three investment plans for this user; Plan 1: Mainly conservative investment, with the proportion of conservative investment being 100%; Option 2: Focus on conservative investments and supplement with aggressive investments, where the proportion of conservative investments is less than 80% but more than 50%. Option 3: Focus on conservative investments and supplement with aggressive investments, where the proportion of conservative investments is more than 80% but less than 100%.

[0017] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for constructing a user profile based on financial transaction data, characterized in that: The following steps are involved: Step 1: Establish a user transaction information database to record the user's historical financial transaction information; Step 2: Preprocess the obtained historical financial transaction information of the user to obtain financial sample data associated with the user; Step 3: Create user profiles based on the obtained financial sample data.

2. The method for constructing a user profile based on financial transaction data according to claim 1, characterized in that: The process of establishing a user transaction information database includes: A registration unit and a login unit are provided in the financial transaction platform for users to register and verify their financial transaction identities, so that users can obtain access to the financial transaction platform; After obtaining the permission to enter the financial transaction platform, a user transaction information database associated with the registered account is established according to the registered account, and the user's financial transaction information in the financial transaction platform is recorded through the user transaction information database.

3. The method for constructing a user profile based on financial transaction data according to claim 2, characterized in that: The process of user identity registration includes: The user's basic personal information is input through the registration unit, and the financial transaction platform reviews the input basic personal information and outputs the review result; After the review is passed, the login account and password will be generated based on the mobile phone number in the personal basic information entered; By inputting the login account and login password into the login unit, the user obtains the authority to enter the financial transaction platform.

4. The method for constructing a user profile based on financial transaction data according to claim 3, characterized in that: The process of preprocessing the obtained user historical financial transaction information includes: Read each transaction record and the transaction amount of each transaction record in the user transaction information database; Obtain the number of overdraft transactions and the overdraft amount of each overdraft transaction, and associate each overdraft amount with the transaction record; Obtain the transaction type corresponding to each transaction record, classify the transaction records according to the transaction type, and obtain a multi-dimensional transaction record data set based on its transaction amount; The multidimensional transaction record data set is normalized so that each column of data in the multidimensional transaction record data set is between [0, 1], and the normalized data set is recorded as financial sample data.

5. The method for constructing a user profile based on financial transaction data according to claim 4, characterized in that: In the process of obtaining transaction records and overdraft transactions, a behavior time window for obtaining transaction records and overdraft transactions is set, and the time of the behavior time window is T, and the time T of the behavior time window is the duration between time t1 and time t2, wherein time t1 is the current time, and time t2 is a past time with a time interval of T from time t1.

6. The method for constructing a user profile based on financial transaction data according to claim 5, characterized in that: The process of user profiling includes: Establish a tree model and use the financial sample data as input into the tree model to obtain the feature set According to the difference between the obtained feature set and the financial sample data, a residual data set is obtained; The obtained residual data set is used as a training set, and training is continuously performed until the obtained residual is 0 or the upper limit of the training times of the tree model is reached, and the final feature set extracted is obtained. The final feature set is marked as a financial feature set, and a user profile is obtained based on the obtained financial feature set.

Images