WO2020177477A1

WO2020177477A1 - Credit service recommendation method, apparatus, and device

Info

Publication number: WO2020177477A1
Application number: PCT/CN2020/070507
Authority: WO
Inventors: 续涛
Original assignee: 阿里巴巴集团控股有限公司
Priority date: 2019-03-07
Filing date: 2020-01-06
Publication date: 2020-09-10
Also published as: CN109903140A

Abstract

Disclosed in embodiments in the present description are a credit service recommendation method, apparatus and device. The solution comprises: acquiring credit performance data to be categorized; inputting the credit performance data to be categorized into a credit service classifier, and outputting the credit service type corresponding to the credit performance data to be categorized; and recommending a credit service according to the credit service type.

Description

Credit service recommendation method, device and equipment

Technical field

This application relates to the field of computer technology, in particular to a credit service recommendation method, device and equipment.

Background technique

With the maturity of the credit evaluation system established around Internet big data, various credit services are becoming more and more available, and users can enjoy the benefits of credit in many areas such as loans, car rentals, house rentals, marriages, and visas. After the user completes the contract fulfillment in various credit application scenarios, the performance data will be generated. The performance data is summarized into the credit account through the orders of each business layer, and is revealed to the user on the client for compliance management display. The current credit account has hundreds of millions of credit performance data, but it lacks a method to analyze the user's performance behaviors through the classification and mining of the performance data, so as to recommend the application scenarios of credit services.

Summary of the invention

In view of this, the embodiments of the present application provide a credit service recommendation method, device, and device, which can analyze the user's performance behavior habits based on credit performance data, thereby recommending credit services.

To solve the above technical problems, the embodiments of this specification are implemented as follows:

A credit service recommendation method provided by an embodiment of this specification includes:

Obtain credit performance data to be classified;

Input the credit performance data to be classified into a credit service classifier, and output the credit service type corresponding to the credit performance data to be classified;

Perform credit service recommendation according to the credit service type.

A credit service recommendation device provided by an embodiment of this specification includes:

The credit performance data acquisition module to be classified is used to obtain the credit performance data to be classified;

A credit service type output module, configured to input the credit performance data to be classified into the credit service classifier, and output the credit service type corresponding to the credit performance data to be classified;

The credit service recommendation module is used to perform credit service recommendation according to the credit service type.

At least one processor; and,

A memory communicatively connected with the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can:

Obtain credit performance data to be classified;

Perform credit service recommendation according to the credit service type.

The above at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects:

The embodiment of this specification determines the credit service type corresponding to the credit performance data to be classified according to the credit service classifier, and then recommends the credit service according to the credit service type. The embodiment of this specification uses the credit service classifier to classify and mine contract performance data, and analyze the user's performance behavior habits, so that the user's contract performance behavior habits can be analyzed based on the credit performance data, so as to recommend credit services.

Description of the drawings

The drawings described here are used to provide a further understanding of the application and constitute a part of the application. The exemplary embodiments and descriptions of the application are used to explain the application and do not constitute an improper limitation of the application. In the attached picture:

FIG. 1 is a schematic flowchart of a credit service recommendation method provided by an embodiment of this specification;

FIG. 2 is a schematic structural diagram of a credit service recommendation device corresponding to FIG. 1 provided by an embodiment of this specification;

FIG. 3 is a schematic structural diagram of a credit service recommendation device corresponding to FIG. 1 provided by an embodiment of this specification.

detailed description

In order to make the purpose, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely in conjunction with specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

The technical solutions provided by the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

FIG. 1 is a schematic flowchart of a credit service recommendation method provided by an embodiment of this specification. From a program perspective, the execution body of the process can be a program or an application client loaded on an application server.

As shown in Figure 1, the process can include the following steps:

Step 101: Obtain credit performance data to be classified.

In the embodiments of this specification, credit performance is data on the performance of credit agreements by credit users in various credit services, which may include data on successful performance or unsuccessful performance. Successful performance refers to the performance of the contract completed by the user within the time limit specified by the credit service. Unsuccessful performance may include the user's failure to complete the performance within the time limit specified by the credit service, that is, the performance beyond the specified time limit, and may also include the user's performance when the time specified by the credit service has not expired.

In the embodiment of this specification, the credit performance data to be classified can be the performance data that the user has completed, it can be one-time credit performance data, or it can be multiple credits for the same (type) credit service within a period of time. Statistics of performance data.

In the embodiment of this specification, the credit performance data to be classified can include business information, such as user ID, merchant ID, business order number, performance record, etc., and can also include user characteristic information, such as user sesame points, user education, and user age , It can also include service provider characteristic information: total number of people in service, total number of people in service, performance rate (probability of user performance).

Step 102: Input the credit performance data to be classified into a credit service classifier, and output the credit service type corresponding to the credit performance data to be classified.

In the embodiment of this specification, the credit service classifier is a model trained based on credit performance data samples, which can extract features of the credit performance data to be classified, and then perform corresponding operations on the credit performance data to be classified according to its internal algorithm Classification, the output result is the credit service type corresponding to the credit performance data to be classified.

In the embodiment of this specification, the biggest difference between the "credit performance data sample" and the "credit performance data to be classified" in step 101 is that the credit service type of the credit performance data sample is known.

In the embodiment of this specification, the user's credit rating is at least one dimension that affects the type of credit service. Among them, the user's credit level is related to the user's credit score and credit performance.

Step 103: Perform credit service recommendation according to the credit service type.

In the embodiment of this specification, after obtaining the credit service type of the credit performance data to be classified, the credit service recommendation can be made for this "credit performance data to be classified".

In the embodiment of this specification, the subject of credit service recommendation may be a user or a credit service provider (merchant). For users, when making credit service recommendations, they can recommend credit services that match the type of credit service. For example, if the credit service type of the credit performance data to be classified is deposit-free, then credit services such as deposit-free car rental and deposit-free umbrella borrowing can be recommended to the user. There can be multiple credit services. For merchants, users who have used the same credit service as the type of service provided by the merchant can be recommended. There can be multiple users, which are displayed in the form of a user list.

The method in Figure 1 determines the type of credit service corresponding to the credit performance data to be classified according to the credit service classifier, and then recommends the credit service according to the type of credit service. The embodiment of this specification uses a credit service classifier to classify and mine contract performance data, and analyze the user's contract performance habits, thereby improving the accuracy of credit service recommendation.

Based on the method in FIG. 1, the embodiments of this specification also provide some specific implementations of the method, which will be described below.

Optionally, before step 101, it may further include:

Obtain samples of credit performance data;

The credit service classifier is trained based on the credit performance data sample, and the credit service type of the credit performance data sample is known.

In the embodiment of this specification, in the embodiment of this specification, the credit performance data sample may be credit performance data of multiple types and multiple channels (user information, merchant information, credit service performance information, external data such as credit reporting data). Input, obtained after data judgment and association.

In the embodiment of this specification, the credit performance data is the performance data generated after the user's performance period under a certain credit service ends. In general, credit performance data may include: user information, merchant information, and credit service performance information. Credit services can include: borrowing umbrellas with no deposit on credit, renting a car with no deposit on credit, trying out credit before buying, and so on. Taking credit deposit-free umbrella borrowing as an example, its credit performance data may include: user information, service provider (merchant) information, borrowing time, borrowing location, return time, and umbrella amount. In addition, the credit performance data sample in this manual may also include some other user information, such as name, certificate type, and certificate number. Moreover, some credit data of a third party can be called to supplement the user's credit information according to the above-mentioned information of the user, which can be a credit investigation system or credit data in other systems, such as credit scores.

In the embodiment of this specification, the number of credit performance data samples can be very large, and can include multiple types of credit performance data. In addition, these credit performance data have been tagged, and the label is used to indicate the credit service type of the credit performance data; that is, the credit service type of each credit performance data sample is known. The embodiment of this specification is to classify the unclassified credit performance data according to the labeled credit performance data, so as to recommend the credit service.

In the embodiment of this specification, the credit service classifier is obtained by training based on multiple types of credit performance data in the credit performance data sample. The classification model can be trained based on the credit big data and the ensemble learning method. It is also possible to make multiple corrections based on the labeled credit performance data until the correct rate of the credit service classifier to classify the credit performance data reaches the preset value. In this way, the more data, the higher the classification accuracy of the credit service classifier. The greater the amount of information in the credit performance data to be classified, the more helpful it is to correctly classify it. The classification accuracy rate is the proportion of correctly classified data in the data set to be classified.

In the embodiment of this specification, the type of credit service may include: credit-free pre-deposit service, credit-free deposit service, credit trial-and-buy service and/or credit reservation service. In addition, for a specific type of credit service, further types can be classified according to the amount of items provided by the credit service. For example, the credit exemption service can be further subdivided into the first credit exemption service and the second credit exemption service. For example, the amount of items provided by the first credit exemption service can be set to be less than 300 yuan, and the second credit exemption service can be set The amount of items provided by the escrow service is more than 300 yuan and less than 800 yuan.

Optionally, before obtaining the credit performance data sample, it may further include:

Acquire first credit performance data whose credit service type is known; the first credit performance data includes user identity information;

Obtain the second credit performance data stored by the third party; the user identity information of the second credit performance data is the same as the user identity information of the first credit performance data;

According to the first credit performance data and the second credit performance data, the credit performance data sample is obtained.

In the embodiment of this specification, the first credit performance data may be data generated and stored during the user's use of the credit service. The stored fields may include user ID, merchant ID, business order number, performance record, etc., the first credit performance Data can be understood as the performance data within the system.

In the embodiments of this specification, the first credit performance data not only includes user identity information, service provider information, and payment transaction information, but also can obtain second credit performance data based on user identity information. The second credit performance data can be understood as the performance data outside the system, that is, the performance data outside the system, such as the performance data information of banks, courts and other institutions. The credit report is one of the manifestations.

In the embodiment of this specification, the user identity information may include: name, certificate type, and certificate number. The certificate types can be ID cards, passports, and driving licenses. The certificate number is the number corresponding to the certificate type. These data can be obtained from the user's registration information. In addition, user information may also include: user credit score, user education, and user age.

In the embodiment of this specification, the credit performance data sample consists of two aspects: the first credit performance data and the second credit performance data, including the user's performance data inside the system and the user's performance data outside the system, which is more conducive to Mining users’ behavior habits and improving the classification accuracy of users’ credit performance data.

Optionally, the acquiring first credit performance data whose credit service type is known may specifically include:

Obtain user information, service provider information, payment information, and/or item information of the performance behavior corresponding to the first credit performance data whose credit service type is known.

In the embodiment of this specification, when there is a service provider in the performance of the contract, the information of the service provider can also be pulled from the credit performance database to further improve the classification accuracy. The "service provider" mentioned here refers to the merchant that provides the service. The information of the service provider may include: business type, total number of people served, total number of people served, and contract performance rate. For example, "Merchant A, deposit-free lease, 128, 247, and 90%" can indicate that the total number of people served by Merchant A is 128, the total number of services is 247, and the user's performance rate is 90%.

In the embodiment of this specification, when payment information exists in the performance of the contract, the transaction-related information can also be pulled from the credit performance database to further improve the accuracy of classification. For example, whether the transaction is successful, whether the transaction has a refund, and the transaction bill. Take the example of renting a power bank with no deposit for credit, 1 yuan per hour, and 2 hours in total, the payment information is 2 yuan.

In the embodiment of this specification, when the performance of the contract includes item information, the item-related information, such as category and price, can also be pulled from the credit performance database to further improve the classification accuracy. Take the credit-free deposit-free rental of power bank as an example. The item category is power bank (daily necessities) and the price is 128 yuan.

In the embodiment of this specification, the credit performance data not only includes the transaction order number and amount field, but may include detailed transaction information to enhance classification. The more specific the information, the more accurate the classification result of the credit service classifier obtained when the credit service classifier is trained later.

Optionally, the training a credit service classifier based on the credit performance data sample may specifically include:

Perform feature extraction on the credit performance data in the credit performance data sample, where the features include: performance-based features, monetary-based features, service provider features, rule features, user features, and/or third-party feedback features;

Vectorize the characterized credit performance data sample;

According to vectorized credit performance data samples, random forest algorithm is used to train multiple credit service type decision trees;

A majority voting principle is adopted for the multiple credit service type decision trees to synthesize a credit service classifier.

In the embodiment of this specification, feature extraction is a feature vector calculated by performing statistics on the behavior of the performance data. According to the fulfillment behavior characteristics and fulfillment scenarios, a matrix is constructed with fulfillment data as row vectors, behavior features as column vectors, and fulfillment scenarios as classification values. The performance scenario is the embodiment of credit service, that is, a specific service provided by a service provider, such as merchant A, XX free deposit borrowing umbrella.

In the embodiment of this specification, the user's credit rating is at least one dimension that affects the type of credit service. User characteristics and third-party feedback characteristics can be two determinants of credit rating. User characteristics may include: user credit score, user education, and user age. The characteristics of third-party feedback can include: whether there is negative information about court arbitration and whether there is a bank breach of contract. For example, the data "user A, 666 points, undergraduate and 32 years old, none" can indicate that user A's credit score is 666 points, the highest degree is undergraduate, the age is 32, and there is no breach of credit in the credit investigation.

In the embodiments of this specification, in addition to the user’s credit rating, the user’s performance behavior habits are another dimension that affects the type of credit service. The user’s performance behavior habits can be mined based on the user’s credit performance data, such as performance characteristics and amount categories. feature. Performance characteristics can include: number/month, whether there is a financial scenario, whether there is a default, and the number of performance scenarios. Amount characteristics can include: performance amount/time, discount amount/time. For example, the data "User C, 16 times/month, financial scenario, no default, 5" indicates that user C has performed 16 times in the month, which is a financial scenario with no default behavior, and there are 5 performance scenarios.

In the embodiments of this specification, in addition to the user's credit level and the user's performance behavior habits, the characteristics of the service provider are also a dimension that affects the type of credit service. The characteristics of the service provider may include: the total number of services, the total number of services, and the performance rate. Rule features can include: service discounts, service access points. The total number of services can indicate how many users the service provider has provided services to. The total number of service visits can indicate how many times the service provider has provided users with services in total. The fulfillment rate can indicate the probability that users who use the service provided by the service provider will successfully fulfill the contract. Service discount can indicate the degree of preferential service provided by the service provider. Such as a 20% discount. The service access score indicates the access threshold for users of the credit service provided by the service provider. For example, users with a credit score of 600 can use the credit service provided by the service provider. For example, "Merchant B, deposit-free lease, 237, 931, 95%, and 650" can indicate that the total number of people served by Merchant B is 237, the total number of services is 931, the user's performance rate is 95%, and the service access It is divided into 650 credit points. From the above data, it can be inferred that the user will use the service provided by merchant B many times, and the user's fulfillment rate is very high.

The main process of the random forest algorithm is as follows:

1) Randomly select N samples from the credit performance database using Bootstrap software.

2) Randomly extract K features from the selected features, and use the C4.5 algorithm to obtain multiple decision trees. Decision tree: A tree structure model induced by top-down recursion of data instances and based on the difference in information entropy. Using the top-down recursive method, the basic idea is to construct a tree with the fastest decrease in entropy value as a measure of information entropy, and the entropy value at the leaf node is 0, that is, the instances of leaf nodes are classified into one category.

3) Repeat the above process until M decision trees, that is, random forests, are obtained, and the performance data are classified according to the voting algorithm to obtain the final model. Random forest uses the idea of ensemble learning to classify the itinerary model of multiple decision trees at the data training place. Ensemble learning is to train multiple classifiers, and finally integrate the classification results to determine the classification idea of the tuple category.

In the embodiment of this specification, the C4.5 algorithm is used as the decision tree algorithm, and the information gain rate is used as the feature split rule to train a set of decision trees. The C4.5 algorithm is a kind of decision tree algorithm. As a classification algorithm, the decision tree algorithm aims to classify n samples with p-dimensional features into c categories. It is equivalent to making a projection, c=f(n), and assigning a class label to the sample after a transformation. In order to achieve this goal, the decision tree can represent the classification process as a tree, and each time it bifurcates by selecting a feature pi.

In the embodiment of this specification, the selected K features may include multiple types of features, such as performance-based features, monetary-based features, user features, and third-party feedback features. It can also be multiple characteristics of the same type of characteristics, such as user characteristics: one or more of user credit score, user education, and user age.

In the embodiment of this specification, the random forest algorithm is adopted, and the method of integrated learning random + voting is used to enhance classification accuracy, resist noise and prevent overfitting, and can obtain high-precision classification accuracy and recall.

In the embodiment of this specification, a random forest algorithm is used to train the credit service classifier. Since the classification accuracy of a single decision tree has large deviations on different classification sets, overfitting may also occur on a single classification set. Overfitting means that the model performs well on the training set, but performs poorly on the test set. The reason is mostly that the selection of the training set is unreasonable. For example, the training set is basically all apples. Using this training set to classify fruits and vegetables, the training model performs particularly well, but the strawberry in the test set cannot be classified. Random forest draws on the idea of ensemble learning, uses sample set sampling, feature set selection, and classification algorithm selection to train different decision trees, and then uses principles such as majority voting to complete the aggregation of results, which can not only improve classification accuracy, but also Effectively avoid overfitting of a single classifier.

Optionally, the performing credit service recommendation according to the credit service type may specifically include:

Determine the credit service that meets the credit service type;

Recommend the credit service to the user corresponding to the credit performance data to be classified.

In the embodiment of this specification, the recommendation to the user may include the following information:

userId: credit service name; top merchants in the scene (configurable); merchants can use stores (online or offline); free deposit amount (exempt amount).

In the embodiment of this specification, merchants with a high fulfillment rate can be recommended to users first. The recommended form can be in the form of "icon + text". Users can click on the corresponding icon to understand the corresponding operating instructions and usage rights.

For example, if the credit service category is "credit-free" service, you can recommend such as credit-free umbrella lending, credit-free deposit-free car rental, credit-free deposit-free rental house, etc. You can also recommend a credit deposit-free service for the corresponding amount based on the user's credit score. For example, a high credit score indicates a high degree of performance, and a relatively large amount of credit free deposit service can be recommended. This is closely related to the item information of the credit performance data obtained in the process of creating the credit performance data sample.

Optionally, the determining the credit service conforming to the credit service type specifically includes:

Determining the user credit level corresponding to the credit service type;

Determine the performance characteristics corresponding to the credit service type;

Select a credit service that meets the user's credit rating and the characteristics of the performance behavior.

In the embodiment of this specification, when determining the credit service that meets the credit service type, the credit service can be screened according to two characteristics, such as the user credit level and contract performance characteristics corresponding to the credit service type.

Take the user's credit rating as an example, because the "credit test first and then buy service" usually requires users to properly keep the trial products, so the user's credit rating is higher. That is, the "try-first-credit-buy service" has higher requirements for the user's credit rating than the "credit-free service" has a higher requirement for the user's credit rating. According to this situation, different credit services can be provided for different user groups. For example, the service access of "Credit first try before buying service" is divided into 650, which means that users are required to have a credit score of 650 or more to enjoy this service, and the service access of "Credit Free Service" is divided into 600, which means the user's credit is required. You can enjoy this service with a score of 600 or more. When the user's credit score is 680, the "try-first-credit-buy service" can be recommended for the user, and when the user's credit score is 610, the "credit-free service" can be recommended for the user.

In addition, credit services can be further screened based on the user's performance characteristics. For example, the credit service type can include a restriction on the amount of the item, and match the amount of the item provided by the credit service. For example, the user’s existing performance behaviors include: renting a car without deposit, renting a bicycle, and the bicycle deposit is 399 yuan, then the corresponding user’s credit service type can be matched to "Credit Free Service, 500", which means that Users can provide deposit-free services for items less than 500 yuan.

Determine the user information corresponding to the credit performance data to be classified;

Determine the service provider of the credit service that meets the credit service type;

Recommend the user information to the service provider.

In the embodiments of this specification, it is also possible to recommend merchants. Since the main body is a merchant, according to the business characteristics of the merchant, users who meet certain conditions need to be pushed to the merchant. The recommended form can be as follows:

Recommendations to merchants can include the following information:

The name of the credit scenario, user A; the store where user A is closest to the merchant; whether user A has performed the credit scenario in the last time; user A has spent the last time in the scenario. User B; the store closest to the merchant; whether user B performed the contract in the last time; user B spent the last time in the scene, and so on.

In the embodiments of this specification, users who meet certain conditions are recommended to merchants based on credit performance data. The merchants can overview the user information and formulate operating strategies based on the user information to improve their services to meet user requirements. There can be multiple users, and the order of the users can be determined according to credit rating, performance rate, etc. Users with high credit ratings and high compliance rates can be recommended first, followed by users with low credit ratings and low compliance rates. Users with default data will not be recommended.

The embodiment of this specification analyzes the user's performance behavior habits by marking and classifying existing data samples, feature extraction, feature vectorization, and classification model training. Based on user performance habits, recommend scenarios for users who have used credit services, and predict and classify users who have not used credit scenarios. For merchants, it can support operation strategy customization and data overview; for users, it can support scene grouping recommendation and scene targeted delivery.

Optionally, after the credit service recommendation is performed according to the credit service type, the method further includes:

Adding the credit performance data corresponding to the credit service to the credit performance data sample.

In the embodiment of this specification, based on the classification model with high accuracy, the user is subjected to directional recommendation and group placement of credit scenarios, and the result of the second use of the recommendation is fed back to the classification model, which further improves the classification accuracy.

Based on the same idea, the embodiment of this specification also provides a device corresponding to the above method. FIG. 2 is a schematic structural diagram of a credit service recommendation device corresponding to FIG. 1 provided by an embodiment of this specification. As shown in Figure 2, the device may include:

The credit performance data acquisition module 201 to be classified is used to obtain the credit performance data to be classified;

The credit service type output module 202 is configured to input the credit performance data to be classified into the credit service classifier, and output the credit service type corresponding to the credit performance data to be classified;

The credit service recommendation module 203 is configured to perform credit service recommendation according to the credit service type.

Optionally, the device may further include:

Credit performance data sample acquisition module for obtaining credit performance data samples;

The credit service classifier training module trains the credit service classifier according to the credit performance data sample, and the credit service type of the credit performance data sample is known.

Optionally, the device may further include:

The first credit performance data acquisition module is used to acquire the first credit performance data whose credit service type is known; the first credit performance data includes user identity information;

The second credit performance data acquisition module is used to acquire second credit performance data stored by a third party; the user identity information of the second credit performance data is the same as the user identity information of the first credit performance data;

The credit performance data sample obtaining module is configured to obtain the credit performance data sample based on the first credit performance data and the second credit performance data.

Optionally, the first credit performance data acquisition module may be specifically used to acquire user information, service provider information, payment information, and/or item information corresponding to the first credit performance data whose credit service type is known .

Optionally, the credit service classifier training module specifically includes:

The feature extraction unit is used to perform feature extraction on credit performance data samples, the features including: performance type features, amount type features, service provider features, rule features, user features, and/or third-party feedback features;

The vectorization unit is used to vectorize the characterized credit performance data sample;

The training unit is used to train multiple credit service type decision trees based on vectorized credit performance data samples using random forest algorithm;

The credit service classifier synthesis unit is used to synthesize the credit service classifier by adopting the majority voting principle for the multiple credit service type decision trees.

Optionally, the credit service recommendation module 203 may specifically include:

A credit service determining unit, configured to determine a credit service that meets the credit service type;

The credit service recommendation unit is configured to recommend the credit service to the user corresponding to the credit performance data to be classified.

Optionally, the credit service determining unit may specifically include:

The user credit level determining subunit is used to determine the user credit level corresponding to the credit service type;

The performance characteristic determination subunit is used to determine the performance characteristic corresponding to the credit service type;

The credit service selection subunit is used to select a credit service that satisfies both the user's credit level and the characteristics of the performance behavior.

A user information determining unit, configured to determine user information corresponding to the credit performance data to be classified;

The service provider determining unit is used to determine the service provider of the credit service conforming to the credit service type;

The user information recommendation unit is configured to recommend the user information to the service provider.

Optionally, the types of credit services include: credit-free pre-deposit service, credit-free deposit service, credit trial-and-buy service, and/or credit reservation service.

Optionally, the device may further include:

The credit performance data adding module is configured to add the credit performance data corresponding to the credit service to the credit performance data sample after the credit service recommendation is performed according to the credit service type.

Technical effects that can be achieved by the embodiments of this specification: omni-directional and multi-channel construction of credit performance data for credit, using random forest algorithm to train the data, and the introduction of integrated learning, reducing the overfitting and non-noise resistance of a single decision tree It can handle discrete and multi-dimensional performance data with high classification accuracy and recall rate.

Based on the same idea, the embodiment of this specification also provides a device corresponding to the above method.

FIG. 3 is a schematic structural diagram of a credit service recommendation device corresponding to FIG. 1 provided by an embodiment of this specification. As shown in FIG. 3, the device 300 may include:

At least one processor 310; and,

A memory 330 communicatively connected with the at least one processor; wherein,

The memory 330 stores instructions 320 executable by the at least one processor 310, and the instructions are executed by the at least one processor 310, so that the at least one processor 310 can:

Obtain credit performance data to be classified;

Perform credit service recommendation according to the credit service type.

The embodiment of this specification first constructs credit performance data samples through multi-channel and multi-type data collection; then randomly selects a certain amount of performance data, and obtains a credit performance data classifier through feature extraction, model training, and integrated learning voting algorithm; then According to the performance data classifier, the performance data of the intended users are classified, and then the credit scenarios are recommended based on the classification results.

In the 1990s, the improvement of a technology can be clearly distinguished between hardware improvements (for example, improvements in circuit structures such as diodes, transistors, switches, etc.) or software improvements (improvements in method flow). However, with the development of technology, the improvement of many methods and processes of today can be regarded as a direct improvement of the hardware circuit structure. Designers almost always get the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be realized by hardware entity modules. For example, a programmable logic device (Programmable Logic Device, PLD) (for example, a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user's programming of the device. It is programmed by the designer to "integrate" a digital system on a PLD without requiring the chip manufacturer to design and manufacture a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly realized by "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code must also be written in a specific programming language, which is called Hardware Description Language (HDL), and there is not only one type of HDL, but many types, such as ABEL (Advanced Boolean Expression Language) , AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description), etc., currently most commonly used It is VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. It should also be clear to those skilled in the art that only a little bit of logic programming of the method flow in the above-mentioned hardware description languages and programming into an integrated circuit can easily obtain the hardware circuit that implements the logic method flow.

The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program codes (such as software or firmware) executable by the (micro)processor. , Logic gates, switches, application specific integrated circuits (ASICs), programmable logic controllers and embedded microcontrollers. Examples of controllers include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicon Labs C8051F320, the memory controller can also be implemented as a part of the memory control logic. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, it is entirely possible to program the method steps to make the controller use logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded The same function can be realized in the form of a microcontroller, etc. Therefore, such a controller can be regarded as a hardware component, and the devices included in it for implementing various functions can also be regarded as a structure within the hardware component. Or even, the device for realizing various functions can be regarded as both a software module for realizing the method and a structure within a hardware component.

The systems, devices, modules, or units illustrated in the above embodiments may be specifically implemented by computer chips or entities, or implemented by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cell 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 Any combination of these devices.

For the convenience of description, when describing the above device, the functions are divided into various units and described separately. Of course, when implementing this application, the functions of each unit can be implemented in the same one or more software and/or hardware.

Those skilled in the art should understand that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention 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.) containing computer-usable program codes.

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, and 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 the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing equipment to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing equipment are generated It is a device that realizes the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing equipment to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including the instruction device. The device implements the functions specified in one process or multiple processes in the flowchart and/or one block or multiple blocks in the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment. The instructions provide steps for implementing functions specified in a flow or multiple flows in the flowchart and/or a block or multiple blocks in the block diagram.

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

The memory may include non-permanent memory in computer readable media, random access memory (RAM) and/or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer readable media.

Computer-readable media include permanent and non-permanent, removable and non-removable media, and information storage can be realized by any method or technology. The 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, CD-ROM, digital versatile disc (DVD) or other optical storage, Magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices. According to the definition in this article, computer-readable media does not include transitory media, such as modulated data signals and carrier waves.

It should also be noted that the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, product or equipment including a series of elements not only includes those elements, but also includes Other elements that are not explicitly listed, or include elements inherent to this process, method, commodity, or equipment. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, method, commodity, or equipment that includes the element.

This 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. This application can also be practiced in distributed computing environments. In these distributed computing environments, remote processing devices connected through a communication network perform tasks. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, as for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and for related parts, please refer to the part of the description of the method embodiment.

The above descriptions are only examples of this application and are not used to limit this application. For those skilled in the art, this application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included in the scope of the claims of this application.

Claims

A credit service recommendation method, including:

Obtain credit performance data to be classified;

Input the credit performance data to be classified into a credit service classifier, and output the credit service type corresponding to the credit performance data to be classified;

Perform credit service recommendation according to the credit service type.
The method according to claim 1, before said obtaining the credit performance data to be classified, further comprising:

Obtain samples of credit performance data;

The credit service classifier is trained based on the credit performance data sample, and the credit service type of the credit performance data sample is known.
The method according to claim 2, before said obtaining the credit performance data sample, further comprising:

Acquire first credit performance data whose credit service type is known; the first credit performance data includes user identity information;

Obtain the second credit performance data stored by the third party; the user identity information of the second credit performance data is the same as the user identity information of the first credit performance data;

According to the first credit performance data and the second credit performance data, the credit performance data sample is obtained.
The method according to claim 3, wherein said obtaining the first credit performance data whose credit service type is known specifically includes:

Obtain user information, service provider information, payment information, and/or item information of the performance behavior corresponding to the first credit performance data whose credit service type is known.
The method according to claim 2, wherein the training a credit service classifier according to the credit performance data sample specifically includes:

Perform feature extraction on credit performance data samples, the features include: performance type features, amount type features, service provider features, rule features, user features, and/or third-party feedback features;

Vectorize the characterized credit performance data sample;

According to vectorized credit performance data samples, random forest algorithm is used to train multiple credit service type decision trees;

A majority voting principle is adopted for the multiple credit service type decision trees to synthesize a credit service classifier.
The method according to claim 1, wherein said performing credit service recommendation according to said credit service type specifically comprises:

Determine the credit service that meets the credit service type;

Recommend the credit service to the user corresponding to the credit performance data to be classified.
The method according to claim 6, wherein the determining the credit service conforming to the credit service type specifically includes:

Determining the user credit level corresponding to the credit service type;

Determine the performance characteristics corresponding to the credit service type;

Select a credit service that meets the user's credit rating and the characteristics of the performance behavior.
The method according to claim 1, wherein said performing credit service recommendation according to said credit service type specifically comprises:

Determine the user information corresponding to the credit performance data to be classified;

Determine the service provider of the credit service that meets the credit service type;

Recommend the user information to the service provider.
The method according to claim 1, wherein the types of credit services include: credit-free pre-deposit service, credit-free deposit service, credit trial-and-buy service, and/or credit reservation service.
The method according to claim 1, after said performing credit service recommendation according to said credit service type, further comprising:

Adding the credit performance data corresponding to the credit service to the credit performance data sample.
A credit service recommendation device, including:

The credit performance data acquisition module to be classified is used to obtain the credit performance data to be classified;

A credit service type output module, configured to input the credit performance data to be classified into the credit service classifier, and output the credit service type corresponding to the credit performance data to be classified;

The credit service recommendation module is used to perform credit service recommendation according to the credit service type.
The apparatus according to claim 11, further comprising:

Credit performance data sample acquisition module for obtaining credit performance data samples;

The credit service classifier training module trains the credit service classifier according to the credit performance data sample, and the credit service type of the credit performance data sample is known.
The device according to claim 12, further comprising:

The first credit performance data acquisition module is used to acquire the first credit performance data whose credit service type is known; the first credit performance data includes user identity information;

The second credit performance data acquisition module is used to acquire second credit performance data stored by a third party; the user identity information of the second credit performance data is the same as the user identity information of the first credit performance data;

The credit performance data sample obtaining module is configured to obtain the credit performance data sample based on the first credit performance data and the second credit performance data.
The device of claim 13, wherein the first credit performance data acquisition module is specifically configured to acquire user information, service provider information, payment information, and performance information corresponding to the first credit performance data whose credit service type is known. / Or item information.
The apparatus according to claim 12, the credit service classifier training module specifically includes:

The feature extraction unit is used to perform feature extraction on credit performance data samples, the features including: performance type features, amount type features, service provider features, rule features, user features, and/or third-party feedback features;

The vectorization unit is used to vectorize the characterized credit performance data sample;

The training unit is used to train multiple credit service type decision trees based on vectorized credit performance data samples using random forest algorithm;

The credit service classifier synthesis unit is used to synthesize the credit service classifier by adopting the majority voting principle for the multiple credit service type decision trees.
The apparatus according to claim 11, wherein the credit service recommendation module specifically comprises:

A credit service determining unit, configured to determine a credit service that meets the credit service type;

The credit service recommendation unit is configured to recommend the credit service to the user corresponding to the credit performance data to be classified.
The apparatus according to claim 11, wherein the credit service recommendation module specifically comprises:

A user information determining unit, configured to determine user information corresponding to the credit performance data to be classified;

The service provider determining unit is used to determine the service provider of the credit service conforming to the credit service type;

The user information recommendation unit is configured to recommend the user information to the service provider.
The apparatus of claim 12, wherein the types of credit services include: credit-free pre-deposit service, credit-free deposit service, credit trial-and-buy service, and/or credit reservation service.
The device according to claim 12, further comprising:

The credit performance data sample adding module is configured to add the credit performance data corresponding to the credit service to the credit performance data sample after the credit service recommendation is performed according to the credit service type.
A credit service recommendation device, including:

At least one processor; and,

A memory communicatively connected with the at least one processor; wherein,

The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor, so that the at least one processor can:

Obtain credit performance data to be classified;

Inputting the credit performance data to be classified into a credit service classifier, and outputting the credit service type corresponding to the credit performance data to be classified;

Perform credit service recommendation according to the credit service type.