CN114820170A

CN114820170A - Customer admission method and device

Info

Publication number: CN114820170A
Application number: CN202210481254.4A
Authority: CN
Inventors: 王建涛; 高凌妤; 程呈
Original assignee: Agricultural Bank of China
Current assignee: Agricultural Bank of China
Priority date: 2022-05-05
Filing date: 2022-05-05
Publication date: 2022-07-29

Abstract

The invention provides a client access method and a client access device, wherein the method comprises the following steps: accessing an ERP system, preprocessing the obtained client pre-loan data of any client transacting any financing product in the ERP system, and obtaining client pre-loan classification data consisting of client basic information, product information and personalized information; verifying the pre-loan classification data of the client; if the client fails to pass the verification, determining that the client fails to be admitted; if the financing products of different types pass the verification, the client evaluation indexes corresponding to the financing products of different types are input into the feature selection model for index selection, the credit scoring indexes of the financing products handled by the clients are output, and the clients are scored based on the credit scoring indexes to obtain client scores; auditing the customers according to the customer scores; if the client passes the audit, the client is determined to be successfully admitted; and if the client fails to pass the audit, determining that the client fails to be admitted. The method determines the client admission condition, improves the service efficiency and ensures the data to be real and effective.

Description

Customer admission method and device

Technical Field

The present invention relates to the field of computer technologies, and in particular, to a method and an apparatus for customer admission.

Background

The medium and small-sized micro enterprises play an important role in national economic development and are essential main force and power sources for national social development, and the credit financing problem of the medium and small-sized micro enterprises becomes a common consensus. Aiming at the financing requirements of small and medium-sized micro enterprises, each bank fully utilizes the characteristics of the supply chain and the enterprises on the chain, and the innovation of supply chain financing products is accelerated. However, the supply chain has complex scene and various financing modes, a single product is difficult to meet the financing requirements of customers, and how to evaluate credit before loan and verify admission of diversified financing customers becomes a big problem.

In the related technology, the credit admission process depends on the off-line financial statement, and the customer manager adjusts the credit admission process as far as possible on the spot, so that the method has low business efficiency and is greatly influenced by human subjective factors. Moreover, the admission conditions of small and medium enterprises in the supply chain do not form a uniform standard, and each financial institution in the market requires different admission data provided by the client, so that information redundancy is easily caused. In addition, the supply chain financial products are based on various financing scenes and financing modes, and the obtained evaluation results of customers are lack of scientificity by using single and same indexes in the existing customer credit scoring model.

Therefore, the existing credit access mode has low service efficiency and redundant information, and can not ensure the reality and the effectiveness of data.

Disclosure of Invention

In view of this, embodiments of the present invention provide a method and an apparatus for customer admission, so as to achieve the purposes of improving service efficiency, avoiding information redundancy, and ensuring data to be true and effective.

In order to achieve the above purpose, the embodiments of the present invention provide the following technical solutions:

the first aspect of the embodiment of the invention discloses a customer admission method, which is suitable for a financing system, and comprises the following steps:

accessing an Enterprise Resource Planning (ERP) system, and acquiring client pre-loan data of any client transacting any financing product in the ERP system;

preprocessing the client pre-loan data to obtain client pre-loan classification data consisting of client basic information, product information and personalized information;

verifying the pre-loan classification data of the client;

if the client fails to pass the verification, determining that the client fails to be admitted;

if the financing products of different types pass the verification, customer evaluation indexes corresponding to the financing products of different types are input into a pre-trained feature selection model DRJMIM for index selection, credit scoring indexes for handling the financing products by the customers are output, and the customers are scored based on the credit scoring indexes to obtain customer scores;

auditing the customers according to the customer scores;

if the client passes the audit, the client is determined to be successfully admitted;

and if the client fails to pass the audit, determining that the client fails to be admitted.

Optionally, the verifying the client pre-loan classification data includes:

checking the basic information of the client in the pre-loan classification data of the client;

if any item of information in the customer basic information does not meet a preset condition, determining that the customer basic information does not pass verification;

and if all items of information in the customer basic information meet preset conditions, determining that the customer basic information passes verification.

Optionally, for different types of financing products, inputting the client evaluation index corresponding to each type of financing product into a pre-trained feature selection model DRJMIM for index selection, and outputting the credit score index for the client to handle the financing product, including:

determining a customer evaluation index corresponding to each type of financing product aiming at different types of financing products;

inputting the customer evaluation indexes corresponding to the financing products of all types into a pre-trained feature selection model DRJMIM;

cleaning the client evaluation indexes corresponding to the financing products of all types to obtain the client evaluation indexes corresponding to the cleaned financing products of all types;

selecting a feature subset from the customer evaluation indexes corresponding to the cleaned financing products of various types, and calculating a weight value of each feature in the feature subset;

selecting the weight of each feature, and updating the weight value of the feature to obtain a feature subset with weight;

outputting a credit score indicator for the customer to transact the financing product based on the weighted feature subset.

Optionally, the process of pre-training the feature selection model DRJMIM includes:

acquiring customer historical data of different types of financing products;

cleaning the historical data of the client to obtain the cleaned historical data of the client;

selecting a feature subset from the washed customer historical data, and assigning an initial weight value to each feature in the feature subset;

setting iteration times, performing iteration according to the iteration times, performing weight selection on each feature, updating the weight value of the feature until the iteration is quitted, and obtaining a result set of each weight value corresponding to the iteration times, wherein the sum of the weight values of all the features is 1;

and weighting and averaging the characteristic values of the characteristics in the result set, and sequencing the average values to obtain the characteristic subset with weight.

Optionally, after obtaining the customer score, the method further includes:

and based on the customer scores, carrying out measurement and calculation by using a pricing test model and a credit calculation model of the financing system to obtain pricing interest rate and credit calculation results of different types of financing products handled by the customers.

Optionally, the auditing the customer according to the customer score includes:

acquiring data information uploaded by the customer after registering the mobile phone number based on the received invitation code, wherein the data information comprises personal user information, loan data information, account information and other information;

respectively carrying out real-name system verification, credit investigation verification, anti-fraud verification, account verification and credit approval verification on the data information and the credit granting result;

if the data information and the credit granting result pass verification, determining that the verification is passed;

and if any one of the data information and the credit granting result is not verified, determining that the verification is not passed.

The second aspect of the embodiment of the invention discloses a customer admission device, which is suitable for a financing system, and comprises:

the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for accessing an Enterprise Resource Planning (ERP) system and acquiring client pre-loan data of any client transacting any financing product in the ERP system;

the preprocessing module is used for preprocessing the client pre-loan data to obtain client pre-loan classification data consisting of client basic information, product information and personalized information;

the verifying module is used for verifying the pre-loan classification data of the client, executing the first determining module if the pre-loan classification data of the client does not pass the verification, and executing the processing module if the pre-loan classification data of the client passes the verification;

the first determining module is used for determining the admission failure of the client;

the processing module is used for inputting the client evaluation indexes corresponding to various types of financing products into a pre-trained feature selection model DRJMIM for index selection, outputting credit scoring indexes for handling the financing products by the client, and scoring the client based on the credit scoring indexes to obtain client scores;

the auditing module is used for auditing the client according to the client score, executing the second confirming module if the client score passes the auditing, and executing the first confirming module if the client score does not pass the auditing;

and the second determining module is used for determining that the client admission is successful.

Optionally, the verification module is specifically configured to:

checking the basic information of the client in the pre-loan classification data of the client; if any one piece of information in the customer basic information does not meet a preset condition, determining that the customer basic information does not pass the verification; and if all items of information in the customer basic information meet preset conditions, determining that the customer basic information passes verification.

Optionally, the processing module includes:

the determining unit is used for determining client evaluation indexes corresponding to financing products of different types aiming at the financing products of different types;

the input unit is used for inputting the client evaluation indexes corresponding to various types of financing products into a pre-trained feature selection model DRJMIM;

the cleaning unit is used for cleaning the client evaluation indexes corresponding to the financing products of all types to obtain the client evaluation indexes corresponding to the financing products of all types after cleaning;

the first processing unit is used for selecting a feature subset from the client evaluation indexes corresponding to the cleaned financing products of various types and calculating a weight value of each feature in the feature subset;

the second processing unit is used for carrying out weight selection on each feature and updating the weight value of the feature to obtain a feature subset with weight;

and the output unit is used for outputting a credit scoring index of the client for transacting the financing product based on the characteristic subset with the weight.

Optionally, the method further includes: a pre-training module;

the pre-training module comprises:

the system comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring client history data of different types of financing products;

the cleaning unit is used for cleaning the client historical data to obtain the cleaned client historical data;

the first processing unit is used for selecting a feature subset from the washed client historical data and assigning an initial weight value to each feature in the feature subset;

the second processing unit is used for setting iteration times, performing iteration according to the iteration times, performing weight selection on each feature, updating the weight value of each feature until the iteration is quitted, and obtaining a result set of each weight value corresponding to the iteration times, wherein the sum of the weight values of all the features is 1;

and the third processing unit is used for weighting and averaging the characteristic values of all the characteristics in the result set and sequencing the average values to obtain the characteristic subset with weight.

Based on the method and the device for customer admission provided by the embodiment of the invention, the method comprises the following steps: accessing an Enterprise Resource Planning (ERP) system, and acquiring client pre-loan data of any client transacting any financing product in the ERP system; preprocessing the client pre-loan data to obtain client pre-loan classification data consisting of client basic information, product information and personalized information; verifying the pre-loan classification data of the client; if the client fails to pass the verification, determining that the client fails to be admitted; if the financing products of different types pass the verification, customer evaluation indexes corresponding to the financing products of different types are input into a pre-trained feature selection model DRJMIM for index selection, credit scoring indexes for handling the financing products by the customers are output, and the customers are scored based on the credit scoring indexes to obtain customer scores; auditing the customers according to the customer scores; if the client passes the audit, the client is determined to be successfully admitted; and if the client fails to pass the audit, determining that the client fails to be admitted. In the scheme, after the client pre-loan classification data passes verification, the client evaluation indexes corresponding to various types of financing products are input to the feature selection model DRJMIM for index selection, the clients are scored based on the output credit scoring indexes, and the clients are audited according to the obtained client scores to determine the client admission condition, so that the service efficiency is improved, information redundancy is avoided, and the data is guaranteed to be real and effective.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.

Fig. 1 is a schematic flowchart of a client admission method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a process for selecting an index according to an embodiment of the present invention;

FIG. 3 is a schematic flow chart of a pre-training feature selection model DRJMIM according to an embodiment of the present invention;

fig. 4 is a flowchart of a DRJMIM algorithm according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating a process of auditing clients according to an embodiment of the present invention;

fig. 6 is an exemplary diagram for checking the data information and the trust result according to the embodiment of the present invention;

fig. 7 is a schematic structural diagram of a client admission apparatus according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In this application, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The terms "first," "second," "third," "fourth," and the like in the description and in the claims of the present application and in the drawings described above, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein.

In order to facilitate understanding of the technical solution of the present invention, technical terms appearing in the present invention are explained:

supply chain finance: the bank manages the fund flow and the logistics of small and medium-sized enterprises in the upstream and downstream around a core enterprise, converts the uncontrollable risk of a single enterprise into the overall controllable risk of a supply chain enterprise, controls the risk to be the lowest financial service by three-dimensionally acquiring various information, realizes the integration of multiple flows of logistics, commodity flow, fund flow, information flow and the like in supply chain finance, solves the problems of difficult financing and guarantee of the upstream and downstream enterprises, can reduce the financing cost of a supply chain by opening the bottleneck of the upstream and downstream financing, and improves the competitiveness of the core enterprise and matched enterprises.

Receivable: in a supply chain financing scenario, for upstream customers, financial products based on accounts receivable as financing bases include forms of warranty financing based on accounts receivable transfer, account receivable pledge financing, bill financing, and the like.

Storage class: in a supply chain financing scenario, upstream and downstream customers are covered, inventory-based mobile financing products, including forms of warehouse financing, future rights, and the like.

Prepayment type: in a supply chain financing scenario, downstream customers are financed based on pre-paid type financing products in a transaction, including forms of order financing, vault financing, and the like.

And (4) business circles: in supply chain financing, a circle is formed around a core enterprise with upstream suppliers and downstream distributors.

And (4) multiple business circles: under the chain financing scene, one customer has a trade relationship with a plurality of core enterprises, and financing can be carried out in a plurality of business circles.

Selecting characteristics: also called Feature subset selection or attribute selection, refers to a process of selecting N features from M features (features) to optimize the indexes of the system, and selecting some features having valid characteristics from the original features to reduce the dimensionality of the data set. Is an important means for improving the performance of the learning algorithm.

DRJMIM: an efficient feature selection algorithm.

Information entropy: entropy is a concept proposed by shannon in 1984 to describe the uncertainty of the source, representing a measure of the amount of information needed to remove the uncertainty, i.e., the amount of information an unknown event may contain. From the information propagation perspective, the information entropy can be expressed as the value of the information, and is used as a standard for measuring the information value.

Data mining: a non-trivial process of obtaining an efficient, novel, potentially useful, and ultimately understandable model from a large amount of data, the broad view of data mining is: from being stored in the database. The process of "mining" useful knowledge from a large amount of data in a data warehouse or other information base. In particular, the huge data is deeply analyzed, valuable hidden information, modes and trends are found out, and reference bases are provided for decision making.

Deep learning: the branch of machine learning is an algorithm which takes an artificial neural network as a framework and performs characterization learning on data.

As known from the background art, the existing credit access mode has low service efficiency and redundant information, and can not ensure the reality and effectiveness of data.

Therefore, in the scheme, after the pre-loan classification data of the client passes verification, the client evaluation indexes corresponding to various types of financing products are input to the feature selection model DRJMIM for index selection, the client is scored based on the output credit scoring indexes, and the client is audited according to the obtained client scoring indexes to determine the client admission condition, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Fig. 1 is a schematic flow chart of a customer admission method according to an embodiment of the present invention, which is applicable to a financing system.

In embodiments of the present invention, financing systems include, but are not limited to, banking financing systems and financial institutions.

The method mainly comprises the following steps:

step S101: accessing an enterprise resource planning ERP system, and acquiring the data before credit of any client transacting any financing product in the ERP system.

It should be noted that Enterprise Resource Planning (ERP) is an Enterprise information management system mainly oriented to the manufacturing industry for integrated management of material resources, capital resources, and information resources.

It will be appreciated that the ERP system stores customer pre-loan data for each customer to transact any financing product.

In the process of implementing step S101 specifically, the financing system accesses the enterprise resource planning ERP system, that is, the financing system docks the enterprise resource planning ERP system, and after the completion of the docking, the financing system obtains the client pre-loan data of any client transacting any financing product in the ERP system.

Step S102: and preprocessing the pre-loan data of the client to obtain the pre-loan classification data of the client consisting of the basic information, the product information and the personalized information of the client.

In the process of implementing step S102 specifically, the obtained client pre-loan data is preprocessed to obtain client pre-loan classification data composed of the client basic information, the product information and the personalized information, that is, the client pre-loan data is classified according to the client basic information, the product information and the personalized information to obtain client pre-loan classification data composed of the client basic information, the product information and the personalized information.

It should be noted that, the financing system interfaces with the enterprise resource planning ERP system to classify the obtained client pre-loan data, and the classification is roughly shown in table 1:

table 1:

step S103: and verifying the pre-loan classification data of the client, and if the pre-loan classification data passes the verification, executing the step S104, and if the pre-loan classification data does not pass the verification, executing the step S107.

It should be noted that, in the embodiment of the present invention, basic customer information in the classified data before the customer is credited is mainly verified, where the verified items include, but are not limited to, data non-null verification, format verification such as certificate numbers and mobile phone numbers, credit system blacklist interception, public security badness information interception, bank restricted person interception, age less than 18 years old or more than 57 years old, and a legal person has an overdue or debt without settlement in the bank, and loan interception for new loan and old loan or for extended loan.

When all the checked items pass, step S104 is performed.

When any of the checked items does not pass, step S107 is performed.

Optionally, the process of executing step 103 to verify the pre-loan classification data of the client mainly includes the following steps:

step S11: and checking the basic information of the client in the classified data before the client credits.

Step S12: and judging whether all items of information in the client basic information meet preset conditions, if so, executing step S13, and if not, executing step S14.

It should be noted that the preset condition includes, but is not limited to, the items related to the verification described above.

In the process of implementing step S12, it is determined whether all items of information in the basic customer information satisfy the preset condition, if yes, it indicates that all items of information in the basic customer information satisfy the preset condition, step S13 is executed, and if no, it indicates that any item of information in the basic customer information does not satisfy the preset condition, step S14 is executed.

Step S13: it is determined that the verification is passed.

In the process of implementing step S13, it is determined that the pre-loan classification data of the customer passes verification on the premise that all items of information in the customer basic information satisfy the preset condition.

Step S14: it is determined that the check is not passed.

In the process of implementing step S14, it is determined that the pre-loan classification data of the client has not been verified on the premise that any one of the pieces of basic client information is determined not to satisfy the preset condition.

Step S104: according to different types of financing products, customer evaluation indexes corresponding to various types of financing products are input to a pre-trained feature selection model DRJMIM for index selection, credit scoring indexes for handling the financing products by customers are output, and the customers are scored based on the credit scoring indexes to obtain customer scores.

It should be noted that, a feature selection model (DRJMIM) is used for selecting indexes of client Information features, quantifying the selected indexes, and optimizing the review process of client Information.

In step S104, different types of financing products include, but are not limited to, financing products corresponding to the receivable class, the advance payment class, the inventory class, and the public class.

That is, the client transacts different financing products, which may be divided into different client evaluation indices according to the receivable class, the advance payment class, the inventory class, and the public class.

The customer evaluation indexes corresponding to the financing products of all types are specifically classified as follows: the customer evaluation index corresponding to the financial product to be collected is shown in table 2, the customer evaluation index corresponding to the financial product to be paid in advance is shown in table 3, the customer evaluation index corresponding to the financial product to be stocked is shown in table 4, and the customer evaluation index corresponding to the financial product to be publicly collected is shown in table 5.

Table 2 (customer evaluation index for financial products to be classified):

table 3 (customer evaluation index for prepaid financial products):

table 4 (customer evaluation index for inventory-type financial product):

serial number	Name of item	Unit of
			1	Storing states
2	Warehouse status
			3	Year of production
4	Goods right person code
			5	Name of right of delivery person
6	Growth rate of nearly one year	％
			7	Annual transaction amount	Wan Yuan
8	Years of collaboration with core enterprises	Year of year
			9	Efficiency of operation
10	Turnover rate	％
			11	Net profit	Wan Yuan
12	Rate of production and sale	％
			13	Load rate	％

Table 5 (customer evaluation index for public financial products):

serial number	Name of item	Unit of
			1	French name
2	Sex of legal person
			3	Age of legal person
4	Company registered capital	Wan Yuan
			5	Company organization code
6	Company location
			7	Company establishment date
8	Nature of the enterprise
			9	Total number of employees
10	The related industries

In the process of implementing step S104 specifically, the feature selection model DRJMIM is trained in advance, for different types of financing products, the customer evaluation index corresponding to each type of financing product is input to the feature selection model DRJMIM as an input of the feature selection model DRJMIM, the customer evaluation index corresponding to each type of financing product is subjected to index selection by using the feature selection model DRJMIM, a credit score index for the customer to handle the financing product is obtained, the credit score index is output, and the customer is scored based on the credit score index, so that a customer score is obtained.

Optionally, in a specific embodiment, the credit score indicator is a credit score model.

That is, the customer is assessed pre-credit using a credit scoring model to obtain a customer score.

Optionally, after obtaining the customer score in step S104, the method further includes:

based on the customer scores, a pricing test model and a credit granting measurement model of the financing system are used for measurement and calculation, and pricing interest rates and credit granting results of different types of financing products handled by customers are obtained.

Step S105: and auditing the customers according to the customer scores, executing step S106 if the customer scores pass the auditing, and executing step S107 if the customer scores do not pass the auditing.

Step S106: and determining that the customer admission is successful.

In the process of implementing step S106, it is determined that the customer is successfully admitted on the premise that it is determined that the customer passes the audit.

Step S107: a customer admission failure is determined.

In the process of implementing step S107 specifically, on the premise that it is determined that the client fails to pass the audit, it is determined that the client is successfully admitted.

According to the customer admission method provided by the embodiment of the invention, the data before credit of any customer handling any financing product in an Enterprise Resource Planning (ERP) system is obtained by accessing the ERP system; preprocessing client pre-loan data to obtain client pre-loan classification data consisting of client basic information, product information and personalized information; verifying the pre-loan classification data of the client; if the client fails to pass the verification, determining that the client fails to be admitted; if the customer evaluation indexes pass the verification, the customer evaluation indexes corresponding to the financing products of different types are input into a pre-trained feature selection model DRJMIM for index selection, credit score indexes for the customer to handle the financing products are output, and the customer is scored based on the credit score indexes to obtain customer scores; auditing the customers according to the customer scores; if the client passes the audit, the client is determined to be successfully admitted; and if the client fails to pass the audit, determining that the client fails to be admitted. In the scheme, after the client pre-loan classification data passes verification, the client evaluation indexes corresponding to various types of financing products are input to the feature selection model DRJMIM for index selection, the clients are scored based on the output credit scoring indexes, and the clients are audited according to the obtained client scores to determine the client admission condition, so that the service efficiency is improved, information redundancy is avoided, and the data is guaranteed to be real and effective.

Based on the customer admission method provided in the embodiment of the present invention, step S104 is executed to input the customer evaluation indexes corresponding to the financing products of different types into the pre-trained feature selection model DRJMIM for index selection, and output the credit score indexes for the customer to handle the financing products, as shown in fig. 2, which is a flow diagram for index selection provided in the embodiment of the present invention, and mainly includes the following steps:

step S201: and determining the customer evaluation index corresponding to each type of financing product aiming at different types of financing products.

For example, for a financing product that should be collected, a customer evaluation index corresponding to the financing product that should be collected is determined.

And determining a client evaluation index corresponding to the pre-payment financing product aiming at the pre-payment financing product.

And determining a client evaluation index corresponding to the inventory financing product aiming at the inventory financing product.

And determining a customer evaluation index corresponding to the public financing product aiming at the public financing product.

Step S202: and inputting the customer evaluation indexes corresponding to the financing products of all types into a pre-trained feature selection model DRJMIM.

In the process of implementing step S202, the feature selection model DRJMIM is trained in advance, and the customer evaluation index corresponding to the financing product to be collected, the customer evaluation index corresponding to the financing product to be paid in advance, the customer evaluation index corresponding to the financing product in stock, and the customer evaluation index corresponding to the financing product in public are input to the feature selection model DRJMIM.

Step S203: and cleaning the client evaluation indexes corresponding to the financing products of all types to obtain the client evaluation indexes corresponding to the financing products of all types after cleaning.

It should be noted that, the amount of pre-loan data of any client transacting any financing product in the obtained ERP system is large, and the situations of complex characteristic values, data redundancy, missing and exception are many, as shown in table 6 (exception data sample table), these have a great influence on the evolution of the subsequent feature selection model DRJMIM, and therefore, data needs to be cleaned, that is: and cleaning the client evaluation indexes corresponding to the financing products of various types.

Table 6:

serial number	Transaction amount (Wanyuan)	Date
			...	...	...
869	Abnormal data	2020/3
			...	...	...
26005	-700 (exception data)	2020/04/26
			...
40992	502	9999/01/01 (Exception data)

The abnormal value processing mode is as follows: data exception conditions include, but are not limited to, data format mismatch and data value exception, and the processing mode is replacement by a median, which is also the most common method in data analysis.

Data missing processing mode: a median is substituted for the missing data value.

Data redundancy processing mode: the data redundant part is deleted, for example: two identical entries, only one reserved.

In the process of implementing step S203 specifically, after the customer evaluation indexes corresponding to each type of financing product are input to the feature selection model DRJMIM, the customer evaluation indexes corresponding to each type of financing product are cleaned by using the feature selection model DRJMIM, so as to obtain the customer evaluation indexes corresponding to each type of financing product after cleaning.

Step S204: and selecting a feature subset from the client evaluation indexes corresponding to the cleaned financing products of various types, and calculating the weight value of each feature in the feature subset.

In the process of implementing step S204 specifically, a feature subset is selected from the customer evaluation indexes corresponding to the cleaned financing products of the respective types, and a weight value of each feature in the feature subset is calculated, that is, a feature subset is selected from data sets of different customers (receivable class, prepaid class, inventory class, and the like), and a weight value of each feature in the selected feature subset is calculated.

Step S205: and selecting the weight of each feature, and updating the weight value of the feature to obtain a feature subset with the weight.

In the process of implementing step S205 specifically, a weight selection is performed on each feature, a final weight is selected, and the weight value of the feature is updated by using the selected weight, so as to obtain a feature subset with a weight.

Step S206: and outputting a credit scoring index of the client for processing the financing product based on the characteristic subset with the weight.

In the process of implementing step S206, a credit score index of the client transacting the financing product is obtained based on the weighted feature subset, and the credit score index of the client transacting the financing product is output.

Based on the customer admission method provided by the embodiment of the invention, after the customer evaluation indexes corresponding to the financing products of all types are determined, the customer evaluation indexes corresponding to the financing products of all types are input to the feature selection model DRJMIM for index selection, and the credit score indexes of the financing products handled by customers are obtained, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Based on the customer admission method provided in the embodiment of the present invention, step S104 is executed to input the customer evaluation indexes corresponding to the financing products of each type into the pre-trained feature selection model DRJMIM for index selection, and output the credit score indexes for the customer to handle the financing products, as shown in fig. 3, which is a schematic flow chart of the pre-trained feature selection model DRJMIM provided in the embodiment of the present invention, and mainly includes the following steps:

step S301: customer history data for different types of financing products is obtained.

In the specific implementation process of step S301, customer history data of the customers handling different types of financing products is obtained from the ERP system.

Step S302: and cleaning the historical data of the client to obtain the cleaned historical data of the client.

In the process of specifically implementing step S302, since the number of the acquired customer history data of different types of financing products is large, and the situations of complex characteristic values, redundant data, missing data, and abnormal data are many, the data needs to be cleaned, that is: and obtaining the client historical data after cleaning.

Step S303: selecting a feature subset from the cleaned customer history data, and assigning an initial weight value to each feature in the feature subset.

In the process of implementing step S303, a feature subset is selected from the washed customer history data of different customers, and an initial weight value is assigned to each feature in the selected feature subset.

Step S304: and setting iteration times, iterating according to the iteration times, selecting the weight of each feature, updating the weight of the feature until iteration is quitted, and obtaining a result set of the iteration times corresponding to each weight.

In step S304, the sum of the weight values of all the features is 1.

Step S305: and weighting and averaging the characteristic values of all the characteristics in the result set, and sequencing the average values to obtain the characteristic subset with weight.

For better understanding of the above, fig. 4 is a flowchart of a DRJMIM algorithm according to an embodiment of the present invention.

First, the feature subset Si, which is respectively filtered from the customer information (receivable class, prepaid class, inventory class, etc.), has a total data set D, and each time the feature is selected, it is equivalent to the reorganization of the customer feature information so that the feature that maximizes the joint mutual information amount I (X, S; Y) is selected. To simplify the calculation, we denote a selected feature in S by Xj, and we translate the goal into finding the feature X K that maximizes the mutual information I (X K, Xj; Y). However, when I (X K, Xj; Y) takes a smaller value, because when joint mutual information is lower, it may mean that there is dependency between features, and not necessarily feature redundancy. Therefore, a weight parameter is introduced, the features are weighted according to the correlation with the selected features, and after a new feature is selected into the feature subset, the dynamic weight is updated, introducing the following formula.

DR(X _i )＝DR(X _i )+C_Ratio(X _i ,X _j )*I(X _j ；Y)。

The following table (table 7) is an explanation of the variables in the formula:

table 7:

secondly, the DRJMIM algorithm overall implementation process mainly comprises the following steps:

step S11: a feature subset S is selected from the data sets D of different customers (receivable class, prepaid class, inventory class, etc.).

Step S12: each feature in the selected customer information subset S is assigned an initial weight value DR.

Step S13: and setting the iteration number K.

Step S14: the iteration begins with a data code algorithm that selects a weight for each feature X K.

Step S15: the weight values of X K are updated, and the sum of the weight values of all features is 1.

Step S16: and exiting iteration to obtain K times of result sets with different weight values.

Step S17: and weighting and averaging the characteristic values of all the characteristics in the result set, and sequencing the average values to obtain the characteristic subset S with the weight.

After training the feature selection model DRJMIM, optimizing the feature selection model DRJMIM, wherein the method comprises the following steps: optimization of parameters of the model and optimization of customer information.

The parameter optimization of the model comprises the following steps:

1. and the iteration times are modified, and for different types of client data, the iteration times can be modified to prevent the over-fitting condition from occurring.

2. And modifying the client characteristic information selection strategy, such as: emphasis is placed on increasing the selected probability of a certain feature value.

The optimization of the customer information comprises the following steps:

1. the more valuable customer characteristic information is recalled, such as: data information of the last three years is more representative than transaction data of five years ago.

2. The data amount types of receivable types and prepayment types have more characteristic values, and errors in the calculation process should be reduced as much as possible.

And after training a feature selection model DRJMIM and optimizing the feature selection model DRJMIM, performing DRJMIM algorithm fusion, wherein the DRJMIM algorithm is mainly used for solving the problems that the indexes of the information features of customers are different in selection and the weights of the indexes are different when the customers access different and multiple products. Examples are as follows:

1. customer admission prepaid product:

when the client A only admits the prepaid product, the system only concerns the client prepaid class index. The relevant index selection for customer a is shown in table 8:

table 8:

index number	Index name	Index weight
			1	Total number of strokes of annual order	W1
...	...	...
			n	Annual transaction amount	Wn

The system comprises a financing system, a client-related data database and a client-related data database, wherein the client-related data database is brought into the financing system, the financing system selects n necessary indexes, the index number is {1, n }, and carries out weight assignment on each index, and when a new client enters a prepaid product, the new client only needs to substitute the client-related data to obtain corresponding indexes and scores.

2. Customer admission pre-payment class, inventory class product:

when customer a has two products admitted simultaneously, the system takes care of the payment-type indicators and the inventory-type indicators, and the relevant indicators for customer B are selected as shown in table 9:

table 9:

the system selects m necessary indexes with index number {1, m }, and carries out weight assignment for each index, when new customer simultaneously accesses the pre-payment class and the stock class data, the new customer only needs to substitute the relevant data of the customer to obtain corresponding index and score.

According to the customer access method provided by the embodiment of the invention, conditions are provided for subsequently selecting indexes by training the feature selection model DRJMIM, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Based on the foregoing customer admission method provided in the embodiment of the present invention, a process of auditing a customer in step S105 is executed, as shown in fig. 5, which is a schematic flow diagram for auditing a customer provided in the embodiment of the present invention, and mainly includes the following steps:

step S501: and acquiring the data information uploaded by the client after registering the mobile phone number based on the received invitation code.

In step S501, the material information includes individual user information, loan material information, account information, and other information.

It should be noted that after the pre-loan classification data of the customer passes the verification, the customer receives the invitation code of the bank, and the customer registers at the client by using a mobile phone number or a mailbox and uploads personal user information, loan data information, account information, other information and the like.

Wherein the individual user information includes: legal names, the place of the house, marital conditions, academic calendars, emergency contact related information and the like, and uploading an identity card image for verification.

The loan data information includes: the Chinese and English full names, the Chinese sign codes, the enterprise scale, the national tax and local tax numbers and the like of the enterprises and uploads a business license.

The customer also needs to select a bank, bind the bank card account and can bind a plurality of bank cards at the same time.

If the customer transacts a plurality of products at different business circles, the customer only needs to register once and selects a payment receiving account for the corresponding business circle and the corresponding product.

In the process of the specific implementation step S501, after the pre-loan classification data of the customer passes the verification, the financing system obtains the information uploaded by the customer after registering the mobile phone number based on the received invitation code.

Step S502: and respectively carrying out real-name system verification, credit investigation verification, anti-fraud verification, account verification and credit approval verification on the data information and the credit granting result.

Fig. 6 is a diagram illustrating an example of verifying the data information and the trust result according to an embodiment of the present invention.

In fig. 6, the internet verification, the real name system verification, the credit investigation verification, the anti-fraud verification, the account verification and the credit authorization verification are performed in sequence.

The credit investigation check comprises but is not limited to 1, the number of times of credit investigation records in the last year exceeds 5, 2, 3 outstanding loans exist in banks, 3, outstanding loans exceed 3, 4, various overdue records exceed 2, 5, the applicant has a loan balance in banks above 5 (including), and 6, the external guarantee balance exceeds 2 million.

The account checking includes but is not limited to 1, account networking checking, account opening person consistency checking, 2, the account is in an abnormal state of non-activation, sleep, freezing and the like, 3, the account is a key concern account, service limitation is realized, and 4, the credit card has outstanding debt.

It should be noted that the above verification and its index parameters can be flexibly selected and configured according to actual needs.

Step S503: and judging whether the data information and the credit granting result pass the verification, if so, executing step S504, and if not, executing step S505.

Step S504: and determining to pass the audit.

In the process of implementing step S504, it is determined that the data information and the trust result both pass verification, and then it is determined that the audit is passed.

Step S505: and determining that the audit is not passed.

In the process of implementing step S505, it is determined that any one of the data information and the trust result fails to pass the verification, and it is further determined that the verification fails.

According to the customer admission method provided by the embodiment of the invention, the customer admission condition is determined by carrying out real-name system check, credit investigation check, anti-fraud check, account check and credit approval check on the data information and the credit granting result, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Corresponding to the customer admission method shown in fig. 1 in the embodiment of the present invention, an embodiment of the present invention further provides a customer admission apparatus, which is suitable for a financing system, as shown in fig. 7, and the customer admission apparatus includes: an acquisition module 71, a pre-processing module 72, a verification module 73, a first determination module 74, a processing module 75, an auditing module 76, and a second determination module 77.

The obtaining module 71 is used for accessing the enterprise resource planning ERP system and obtaining the client pre-loan data of any financing product transacted by any client in the ERP system.

And the preprocessing module 72 is used for preprocessing the client pre-loan data to obtain client pre-loan classification data consisting of the client basic information, the product information and the personalized information.

And the checking module 73 is used for checking the classified data before the client credits, executing the first determining module if the classified data does not pass the checking, and executing the processing module if the classified data passes the checking.

A first determining module 74 for determining customer admission failures.

The processing module 75 is configured to, for different types of financing products, input the customer evaluation index corresponding to each type of financing product into the pre-trained feature selection model DRJMIM for index selection, output a credit score index for the customer to handle the financing product, and score the customer based on the credit score index to obtain a customer score.

And the auditing module 76 is used for auditing the customers according to the customer scores, executing the second confirming module 77 if the audits are passed, and executing the first confirming module 74 if the audits are not passed.

A second determination module 77 for determining that the customer admission was successful.

Optionally, based on the customer admission apparatus shown in fig. 7, the checking module 73 is specifically configured to:

checking the basic information of the client in the pre-loan classification data of the client; if any item of information in the customer basic information does not meet the preset condition, determining that the customer basic information does not pass the verification; and if all items of information in the basic information of the client meet the preset conditions, determining that the items of information pass the verification.

Optionally, based on the customer admission apparatus shown in fig. 7, the processing module 75 is configured to score the customer based on the credit score index, and after obtaining the customer score, further configured to:

It should be noted that, the specific principle and the implementation process of each module in the client admission apparatus disclosed in the embodiment of the present invention are the same as those of the client admission method implemented in the present invention, and reference may be made to corresponding parts in the client admission method disclosed in the embodiment of the present invention, which are not described herein again.

According to the customer access device provided by the embodiment of the invention, the Enterprise Resource Planning (ERP) system is accessed to obtain the customer pre-loan data of any customer handling any financing product in the ERP system; preprocessing client pre-loan data to obtain client pre-loan classification data consisting of client basic information, product information and personalized information; verifying the pre-loan classification data of the client; if the client fails to pass the verification, determining that the client fails to be admitted; if the customer evaluation indexes are verified, the customer evaluation indexes corresponding to the financing products of different types are input into a pre-trained feature selection model DRJMIM for index selection, credit score indexes for the customer to handle the financing products are output, and the customer is scored based on the credit score indexes to obtain customer scores; auditing the customers according to the customer scores; if the client passes the audit, the client is determined to be successfully admitted; and if the client fails to pass the audit, determining that the client fails to admit. In the scheme, after the client pre-loan classification data passes verification, the client evaluation indexes corresponding to various types of financing products are input to the feature selection model DRJMIM for index selection, the clients are scored based on the output credit scoring indexes, and the clients are audited according to the obtained client scores to determine the client admission condition, so that the service efficiency is improved, information redundancy is avoided, and the data is guaranteed to be real and effective.

Optionally, based on the customer admission apparatus shown in fig. 7, the processing module 75 includes:

and the determining unit is used for determining the client evaluation index corresponding to each type of financing product aiming at different types of financing products.

And the input unit is used for inputting the client evaluation indexes corresponding to various types of financing products into the pre-trained feature selection model DRJMIM.

And the cleaning unit is used for cleaning the client evaluation indexes corresponding to the financing products of all types to obtain the client evaluation indexes corresponding to the financing products of all types after cleaning.

And the first processing unit is used for selecting the characteristic subset from the client evaluation indexes corresponding to the cleaned financing products of various types and calculating the weight value of each characteristic in the characteristic subset.

And the second processing unit is used for selecting the weight of each feature and updating the weight value of the feature to obtain a feature subset with the weight.

And the output unit is used for outputting the credit scoring index of the financing product transacted by the client based on the characteristic subset with the weight.

Based on the customer access device provided by the embodiment of the invention, after the customer evaluation indexes corresponding to the financing products of various types are determined, the customer evaluation indexes corresponding to the financing products of various types are input to the feature selection model DRJMIM for index selection, and the credit score indexes of the financing products handled by customers are obtained, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Optionally, based on the client admission apparatus shown in fig. 7, in combination with fig. 7, the client admission apparatus further includes a pre-training module 78, where the pre-training module 78 includes:

the acquisition unit is used for acquiring the client history data of different types of financing products.

The cleaning unit is used for cleaning the historical data of the client to obtain the cleaned historical data of the client;

and the first processing unit is used for selecting a characteristic subset from the washed customer historical data and assigning an initial weight value to each characteristic in the characteristic subset.

And the second processing unit is used for setting iteration times, performing iteration according to the iteration times, performing weight selection on each feature, and updating the weight value of the feature until the iteration is quitted to obtain a result set of each weight value corresponding to the iteration times.

Wherein, the sum of the weight values of all the characteristics is 1.

According to the customer access device provided by the embodiment of the invention, conditions are provided for subsequently selecting indexes by training the feature selection model DRJMIM, so that the service efficiency is improved, the information redundancy is avoided, and the data is ensured to be real and effective.

Optionally, based on the client admission apparatus shown in fig. 7, the auditing module 76 is specifically configured to:

acquiring data information uploaded by a client after registering a mobile phone number based on a received invitation code, wherein the data information comprises personal user information, loan data information, account information and other information; respectively carrying out real-name system verification, credit investigation verification, anti-fraud verification, account verification and credit approval verification on the data information and the credit granting result; if the data information and the credit granting result pass verification, determining that the verification is passed; and if any one of the data information and the credit granting result is not verified, determining that the verification is not passed.

According to the customer access device provided by the embodiment of the invention, the customer access condition is determined by carrying out real-name system check, credit investigation check, anti-fraud check, account check and credit approval check on the data information and the credit granting result, so that the service efficiency is improved, the information redundancy is avoided, and the data is real and effective.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the system or system embodiments, which are substantially similar to the method embodiments, are described in a relatively simple manner, and reference may be made to some descriptions of the method embodiments for relevant points. The above-described system and system embodiments are only illustrative, wherein the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for customer admission, adapted for use in a financing system, the method comprising:

verifying the pre-loan classification data of the client;

auditing the customers according to the customer scores;

2. The method of claim 1, wherein said verifying said customer pre-loan classification data comprises:

3. The method according to claim 1, wherein the step of inputting customer evaluation indexes corresponding to different types of financing products into a pre-trained feature selection model DRJMIM for index selection and outputting credit score indexes for handling the financing products by the customer comprises the following steps:

inputting customer evaluation indexes corresponding to various types of financing products into a pre-trained feature selection model DRJMIM;

4. The method of claim 1, wherein the pre-training of the feature selection model DRJMIM comprises:

acquiring client historical data of financing products of different types;

5. The method of claim 1, further comprising, after said obtaining a customer score:

6. The method of claim 1, wherein said auditing said customer based on said customer score comprises:

7. A customer admission apparatus adapted for use in a financing system, the apparatus comprising:

the processing module is used for inputting customer evaluation indexes corresponding to various types of financing products into a pre-trained feature selection model DRJMIM for index selection, outputting credit score indexes for handling the financing products by customers, and scoring the customers based on the credit score indexes to obtain customer scores;

8. The apparatus of claim 7, wherein the verification module is specifically configured to:

checking the basic information of the client in the pre-loan classification data of the client; if any item of information in the customer basic information does not meet a preset condition, determining that the customer basic information does not pass verification; and if all items of information in the customer basic information meet preset conditions, determining that the customer basic information passes verification.

9. The apparatus of claim 7, wherein the processing module comprises:

10. The apparatus of claim 7, further comprising: a pre-training module;

the pre-training module comprises:

the first processing unit is used for selecting a feature subset from the cleaned customer historical data and assigning an initial weight value to each feature in the feature subset;