Novel agricultural operation subject credit granting and credit increasing method, device and equipment and storage medium

Abstract

The invention provides a novel agricultural operation subject credit granting and increasing method and device, equipment and a storage medium. The method comprises the steps that multiple participants acquire data sets related to all novel agricultural operation subjects from respective data sources; the data set comprises related data of functional departments, related data from the novel agricultural operation subject information management platform and related data from banks; performing longitudinal entity alignment association on the data set by using a unique identifier of a novel agricultural operation subject; inputting the data set related to the novel agricultural operation subject into a credit line prediction model of the novel agricultural operation subject based on privacy calculation to obtain a credit line; and determining a target credit line of the novel agricultural operation subject based on the credit line. Through real-time convergence and fusion calculation of multi-channel information, the information transparency between a bank and a novel agricultural operation subject is improved, the credit line of the novel agricultural operation subject is improved by using multi-source data, and the method is high in efficiency, low in cost, objective, accurate and sustainable.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a new type of agricultural business entity credit enhancement method, device, equipment and storage medium. Background Art

[0002] At present, banks' collection of information on new agricultural operators mainly relies on self-reporting by new agricultural operators and offline manual investigations by banks. The data obtained is limited in scope, subjective, not sustainable, and cannot be reused. The workload is large, and the manpower, time, and financial costs are high, which affects the efficiency of credit approval and the accuracy of credit limit calculation. Information asymmetry has resulted in the loan amount available to new agricultural operators being far from meeting the needs of production expansion, which has greatly hindered the sustainable development of the agricultural industry. Summary of the invention

[0003] The present invention provides a method, device, equipment and storage medium for credit enhancement of new agricultural business entities, which are used to solve the defects in the prior art that the current bank's information collection scope and quantity on new agricultural business entities are limited, which affects the efficiency of credit approval and the accuracy of credit limit calculation, resulting in the loan amount available to the new agricultural business entities being insufficient to meet the needs of production expansion.

[0004] The present invention provides a new type of credit enhancement method for agricultural business entities, comprising the following steps: Multiple participants obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from the new agricultural business entity information management platform, and third relevant data from banks; the data sets are vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; Inputting the data set related to the new agricultural business entity into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit enhancement line; Determine the target credit limit of the new agricultural business entity based on the credit enhancement limit; Among them, the default prediction model in the credit limit prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers. The local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data has defaulted, and the ratio of the training data set to the test data set is 7:3.

[0005] According to a novel credit enhancement method for agricultural business entities provided by the present invention, the credit limit prediction model includes the default prediction model and a credit enhancement limit calculation module for full-process multi-party collaboration; The default prediction model is used to: Inputting the data sets related to each new agricultural business entity into the default prediction model, obtaining the predicted value of the default rate of the new agricultural business entity and the importance of each feature within the default prediction model, and dividing the importance of each feature into a functional department data feature importance set, a new agricultural business entity information management platform data feature importance set and a bank data feature importance set according to different sources; The credit enhancement limit calculation module is used to: Determine a first credit enhancement limit based on the predicted default rate value and the functional department characteristic data and the characteristic importance corresponding to the functional department characteristic data; determine a second credit enhancement limit based on the predicted default rate value, the new agricultural business entity information management platform characteristic data and the characteristic importance corresponding to the new agricultural business entity information management platform characteristic data; and determine a third credit enhancement limit based on the predicted default rate value, the bank characteristic data and the characteristic importance corresponding to the bank characteristic data; and determine the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit; The functional department characteristic data are the characteristic data of new agricultural business entities from the functional departments; The new agricultural business entity information management platform characteristic data is the new agricultural business entity characteristic data from the new agricultural business entity information management platform; The bank characteristic data are characteristic data of new agricultural business entities from banks.

[0006] According to a new type of credit enhancement method for an agricultural business entity provided by the present invention, the credit enhancement limit is determined based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit, including: Determine the credit enhancement base and credit enhancement ceiling set by the bank for a specific batch of new agricultural business entities, as well as the additional credit enhancement quota given by the bank to a specific credit group; Determining a non-default rate based on the predicted default rate value; The credit enhancement limit is determined based on the first credit enhancement limit, the second credit enhancement limit, the third credit enhancement limit, the credit enhancement upper limit, the additional credit enhancement limit, the non-default rate, and the credit enhancement base.

[0007] According to a novel credit enhancement method for agricultural business entities provided by the present invention, the credit limit prediction model further includes a vertical federated learning entity alignment module; The vertical federated learning entity alignment module is used to perform vertical federated learning entity alignment on the data sets related to each new agricultural business entity, and then input them into the credit limit prediction model of the new agricultural business entity based on privacy computing.

[0008] According to a new type of agricultural business entity credit enhancement method provided by the present invention, the steps of obtaining the functional department characteristic data, the new type of agricultural business entity information management platform characteristic data and the bank characteristic data include: Performing data preprocessing and feature engineering on the first relevant data to obtain the functional department feature data; Performing data preprocessing and feature engineering on the second relevant data to obtain feature data of the new agricultural business entity information management platform; Performing data preprocessing and feature engineering on the third related data to obtain the bank feature data; The data preprocessing includes at least two of data cleaning, removing outliers, removing duplicate values, removing erroneous values, filling missing values, standardization and normalization; The feature engineering includes at least two of feature extraction, feature dimensionality reduction and feature balancing.

[0009] According to a new type of agricultural business entity credit enhancement method provided by the present invention, the target credit limit of the new type of agricultural business entity is determined based on the credit enhancement limit, including: The target credit limit is determined based on the original credit limit and the credit enhancement limit.

[0010] The present invention also provides a new type of credit enhancement device for agricultural business entities, comprising the following units: An acquisition unit is used for multiple participants to respectively acquire data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from the new agricultural business entity information management platform, and third relevant data from banks; the data sets are vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; An input unit, used to input the data set related to the new agricultural business entity into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit line of the credit enhancement; A determination unit, configured to determine a target credit limit of the new agricultural business entity based on the credit enhancement limit; Among them, the default prediction model in the credit limit prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers. The local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and the subject label default rate and whether the data corresponds to the default label. The ratio of the training data set to the test data set is 7:3.

[0011] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for credit enhancement for a new agricultural business entity as described above is implemented.

[0012] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the credit enhancement method for the new agricultural business entity as described in any one of the above-mentioned methods.

[0013] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for credit enhancement for new agricultural business entities.

[0014] The invention provides a method, device, equipment and storage medium for credit enhancement of new agricultural business entities. Multiple participants obtain data sets related to each new agricultural business entity from their respective data sources. The data sets include first relevant data from functional departments, second relevant data from the information management platform of new agricultural business entities and third relevant data from banks. The data sets are vertically aligned and associated with entities using the unique identifier of the new agricultural business entity. The data sets related to the new agricultural business entity are then input into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit enhancement line. Finally, based on the credit enhancement line, the target credit line of the new agricultural business entity is determined. This process improves the information transparency between banks and new agricultural business entities through real-time convergence and fusion calculation of multi-channel information, and uses multi-source data to improve the credit line of new agricultural business entities. The method is efficient, low-cost, objective and accurate, and the method is sustainable. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is one of the flow charts of the new agricultural business entity credit enhancement method provided by the present invention.

[0017] Figure 2 This is the second flow chart of the new agricultural business entity credit enhancement method provided by the present invention.

[0018] Figure 3 It is a flow chart of the credit enhancement process for the new agricultural business entity provided by the present invention.

[0019] Figure 4 It is a structural schematic diagram of a new agricultural business entity credit enhancement device provided by the present invention.

[0020] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, rather than to describe a specific order or precedence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type.

[0023] Figure 1 This is one of the flow charts of the new agricultural business entity credit enhancement method provided by the present invention, such as Figure 1 As shown, the method includes step 110 , step 120 and step 130 .

[0024] Step 110, multiple participants respectively obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from the new agricultural business entity information management platform and third relevant data from banks; the data sets are vertically aligned with the unique identifier of the new agricultural business entity.

[0025] Specifically, multiple participants obtain data sets related to each new agricultural business entity from their respective data sources, where the data sets include first relevant data from functional departments, second relevant data from the new agricultural business entity information management platform, and third relevant data from banks.

[0026] The first relevant data from the functional departments include: the public social credit score of new agricultural operators, the credit score within the agricultural industry, agricultural subsidy data over the years, agricultural production data, sales data, and land information. Among them, the public social credit score and the credit score within the agricultural industry are calculated by the functional departments according to the rules, and the score range is 35-95 points. The subsidy data over the years include: vegetable field subsidies, soil fertility subsidies, agricultural machinery subsidies and other data.

[0027] The second relevant data from the new agricultural business entity information management platform includes tax payment certificates, financial statements, asset certificates, and Sesame Credit scores of the legal person / person in charge on the third-party payment platform.

[0028] The third-party related data from banks include loan records, credit ratings, and deposit account transaction details of new agricultural business entities. Among them, loan records include loan time, loan amount, interest rate, repayment date, repayment amount, number of defaults, default date, and default amount.

[0029] The dataset is vertically aligned with the unique identifier of new agricultural business entities. The unique identifier of family farms, large-scale growers, agricultural cooperatives, leading agricultural enterprises, and agricultural social service organizations is the unified social credit code. For new agricultural business entities that involve legal person / person in charge information, the ID card is used as the unique identifier.

[0030] Step 120: Input the data set related to the new agricultural business entity into the credit limit prediction model of the new agricultural business entity based on privacy computing to obtain the credit enhancement limit.

[0031] Specifically, the data set related to the new agricultural business entity can be input into the credit limit prediction model of the new agricultural business entity based on privacy computing to obtain the credit enhancement limit.

[0032] Here, the credit limit prediction model for new agricultural operators based on privacy computing includes: a vertical federated learning entity alignment module, a decentralized federated learning model (default prediction model), homomorphic encryption transmission, and a full-process multi-party collaborative credit enhancement limit calculation module.

[0033] Step 130, determining a target credit limit of the new agricultural business entity based on the credit enhancement limit; Among them, the default prediction model in the credit limit prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers. The local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data has defaulted, and the ratio of the training data set to the test data set is 7:3.

[0034] Specifically, after obtaining the credit enhancement line, the target credit line of the new agricultural business entity can be determined based on the credit enhancement line.

[0035] Among them, the default prediction model in the credit limit prediction model is obtained after each participant server is trained locally based on local training data, and then the training parameters of all participant servers are used for global coordination and aggregation. The default prediction model is a decentralized federated learning model.

[0036] Decentralized federated learning model: Federated learning protects the privacy of data during transmission and storage. Each participant encrypts the data locally, and then uses federated learning to train the model, ensuring data privacy while achieving collaboration. Each participant only knows its own characteristics and split points, and the data is only visible to itself and will not be leaked to other participants. The default prediction model construction process includes downloading the global model, updating the model based on local data, uploading and updating after encryption, and server aggregation update. During the local model training process, the model parameters are encrypted. Banks are both participants and collaborators. In addition to providing data, banks are also responsible for key generation, distribution and data aggregation calculation. Bank data contains labels on whether new agricultural operators have defaulted. The functional department management platform and the new agricultural operator information management platform are participants, and are mainly responsible for providing data.

[0037] The construction process of the decentralized federated learning model: 1. Preliminary construction of the global model for default prediction: The global model and its initial parameters are initialized, and then the global model is shared with all participants in the federated learning environment. 2. Training of the local model for default prediction: After sharing the initial model and parameters with all participants, the initial model is trained locally using the training data, and each participant encrypts their own data using additive homomorphic encryption technology. 3. Aggregation of the default prediction model: Participants train local models, perform aggregation calculations based on feature splitting points, and send the results and parameter updates to the collaborating party server for aggregation and training of the global model. The collaborating party merges the models of each participant into a global model, the global model is updated, and the improved model is shared among the participants for the next iteration until the global model prediction accuracy reaches the expected level and the federated learning model is built.

[0038] Homomorphic encryption transmission: Homomorphic encryption protects the privacy of data during joint computing. To prevent the gradients or parameter updates of the model from being intercepted during transmission, homomorphic encryption is used to encrypt these updates to protect the privacy of model updates during transmission. Homomorphic encryption technology is used to protect label information, and key tree enhancement steps are performed on the ciphertext. In the case of multi-party collaboration, the security of the data is ensured, and the collaborating server can securely aggregate them in an encrypted state without decryption. The Paillier addition semi-homomorphic encryption algorithm is adopted here, which is not specifically limited in the embodiments of the present invention.

[0039] Here, the training data set and the test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data is in default, and the ratio of the training data set to the test data set is 7:3.

[0040] It should be noted that the method provided by the embodiment of the present invention: 1) collects credit-related data of new agricultural business entities through multiple channels across platforms, enriches the source of credit data, and the data can be mutually verified. Compared with the data collected by a bank through a single channel, the data has a wider range and more data items, and the basis for credit calculation is more sufficient and the evaluation result is more accurate, which greatly reduces the cost of banks' independent data collection; 2) Homomorphic encryption and federated learning are used to collaboratively build a trust enhancement model for the three parties. The data does not leave the domain, providing comprehensive protection for the data from source to destination, enhancing privacy protection and data processing capabilities, and innovating the way of data fusion, utilization, and data sharing.

[0041] In the method provided by the embodiment of the present invention, multiple participants obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from the information management platform of the new agricultural business entity, and third relevant data from banks; the data sets are vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; the data sets related to the new agricultural business entity are then input into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit line increase; finally, based on the credit line increase, the target credit line of the new agricultural business entity is determined. This process improves the information transparency between banks and new agricultural business entities through real-time convergence and fusion calculation of multi-channel information, and uses multi-source data to increase the credit line of new agricultural business entities. The method is efficient, low-cost, objective, accurate, and sustainable.

[0042] Based on the above embodiment, the credit limit prediction model includes the default prediction model and a credit enhancement limit calculation module for full-process multi-party collaboration; The default prediction model is used to: Inputting the data sets related to each new agricultural business entity into the default prediction model, obtaining the predicted value of the default rate of the new agricultural business entity and the importance of each feature within the default prediction model, and dividing the importance of each feature into a functional department data feature importance set, a new agricultural business entity information management platform data feature importance set and a bank data feature importance set according to different sources; The credit enhancement limit calculation module is used to: Determine a first credit enhancement limit based on the predicted default rate value and the functional department characteristic data and the characteristic importance corresponding to the functional department characteristic data; determine a second credit enhancement limit based on the predicted default rate value, the new agricultural business entity information management platform characteristic data and the characteristic importance corresponding to the new agricultural business entity information management platform characteristic data; and determine a third credit enhancement limit based on the predicted default rate value, the bank characteristic data and the characteristic importance corresponding to the bank characteristic data; and determine the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit; The functional department characteristic data are the characteristic data of new agricultural business entities from the functional departments; The new agricultural business entity information management platform characteristic data is the new agricultural business entity characteristic data from the new agricultural business entity information management platform; The bank characteristic data are characteristic data of new agricultural business entities from banks.

[0043] Specifically, the credit limit prediction model includes a default prediction model and a credit enhancement limit calculation module that involves multi-party collaboration throughout the entire process.

[0044] XGBoost performs well in processing large-scale data sets, has efficient training speed and prediction performance, and supports parallel processing and missing value processing. Since SecureBoost inherits the efficiency and powerful modeling capabilities of XGBoost, and uses privacy protection technology to make it suitable for scenarios with higher data privacy requirements, SecureBoost is used here to build a default prediction model.

[0045] Among them, the default prediction model is used to: Input the data sets related to each new agricultural business entity into the default prediction model to obtain the default rate prediction value p of the new agricultural business entity output by the default prediction model and the importance of each feature within the default prediction model Then, according to different sources, the importance of each feature is divided into the importance set of functional department data features, the importance set of new agricultural business entity information management platform data features, and the importance set of bank data features. Among them, the importance of the feature indicates the degree of influence of each feature on the predicted value of the default rate.

[0046] Specifically, the initial default prediction model can be trained using the functional department management platform, the new agricultural business entity information management platform, the bank's third-party data, and the loan default records of existing new agricultural business entities in the bank as sample data.

[0047] Here, the credit enhancement limit calculation module is used to determine the first credit enhancement limit based on the default rate prediction value and the functional department characteristic data and the characteristic importance corresponding to the functional department characteristic data, determine the second credit enhancement limit based on the default rate prediction value, the new agricultural business entity information management platform characteristic data and the characteristic importance corresponding to the new agricultural business entity information management platform characteristic data, and determine the third credit enhancement limit based on the default rate prediction value, the bank characteristic data and the characteristic importance corresponding to the bank characteristic data, and determine the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit.

[0048] Among them, the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit are all determined based on the following formula:

[0049] in, , and Respectively represent the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit, Indicates the credit enhancement base. represents the predicted value of default rate, , and They represent the normalized functional department characteristic data, new agricultural business entity information management platform characteristic data and bank characteristic data respectively. , and They respectively represent the importance of the characteristics corresponding to the characteristic data of functional departments, the importance of the characteristics corresponding to the characteristic data of the information management platform for new agricultural business entities, and the importance of the characteristics corresponding to the characteristic data of banks. m, n, and t are the number of new agricultural business entity characteristics contained locally by the three participants, namely, functional departments, information management platform for new agricultural business entities, and banks.

[0050] , and From each feature data x 1j , x 2j , x 3j After minimum-maximum normalization, the formula is as follows:

[0051] in, Represents each feature data x 11 , x 12 , …, x 1m ;x 21 , x 22 , …, x 2n ;x 31 , x 32 , …, x 3t , A complete dataset representing a certain characteristic of new agricultural business entities.

[0052] In summary, the above formula can be simplified as follows: normalize each feature data, multiply it by the corresponding feature importance, then multiply it by the probability of non-default to get the credit enhancement rate, and finally multiply it by the credit enhancement base set by the bank. , get credit enhancement line .

[0053] Based on the above embodiment, determining the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit, and the third credit enhancement limit includes: Step 210, determining the credit enhancement base and credit enhancement ceiling set by the bank for a specific batch of new agricultural business entities, as well as the additional credit enhancement limit given by the bank to a specific credit group; Step 220, determining a non-default rate based on the predicted default rate value; Step 230, determining the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit, the third credit enhancement limit, the credit enhancement upper limit, the additional credit enhancement limit, the non-default rate, and the credit enhancement base.

[0054] Specifically, determine the upper limit of credit enhancement set by the bank for a specific batch of new agricultural business entities , and additional credit enhancement lines granted by banks to specific credit groups, such as credit villages .

[0055] The default rate prediction value is , then the non-default rate is .

[0056] Then, based on the first credit enhancement limit, the second credit enhancement limit, the third credit enhancement limit, the credit enhancement limit, the additional credit enhancement limit, and the non-default rate, the formula for determining the credit enhancement limit is as follows:

[0057] in, Indicates the credit enhancement limit. , and Respectively represent the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit, represents the predicted value of default rate, Indicates additional credit enhancement amount. Indicates the upper limit of credit enhancement. Generally, the credit enhancement limit is set as the basic unit of credit enhancement. Integer multiples of Round down to get the final credit enhancement limit .

[0058] in, To accumulate the credit enhancement limit of three parties, that is, to obtain the aggregated credit enhancement limit.

[0059] Update of credit line for new agricultural business entities within the bank: 1. Input the processed data into the privacy calculation model of the credit enhancement line of the new agricultural business entity to obtain the credit enhancement line .

[0060] 2. The original credit limit calculated based on the bank's internal model , and the combined calculation results in the new target credit limit for new agricultural operators. .

[0061] The formula for the target credit limit is as follows:

[0062] If the new agricultural business entity currently has no credit record, that is, it has no loan record, credit score, credit limit or other credit-related information in the bank, the original credit limit will be set to 0.

[0063] The new credit line replaces the original credit line and is stored locally in the bank for new agricultural operators to use as a reference when applying for loans.

[0064] Data update mechanism: There is a data interface between the functional department management platform, the new agricultural business entity information management platform, and the bank and new agricultural business entity credit enhancement limit privacy calculation model server. Once the data is updated, it will be pushed to the server in real time for incremental updates. The more data is gathered, the more objective the credit limit calculation will be.

[0065] In order to avoid frequent changes in credit limits caused by frequent data changes, which would waste computing power, the credit limits are updated by banks actively pulling them. When a new agricultural operator applies for a loan, the bank actively accesses the credit enhancement credit limit privacy computing model server for the new agricultural operator, executes the credit limit calculation program, and obtains the credit enhancement credit limit. Then, the credit limit of the new agricultural operator within the bank is updated according to the rules.

[0066] The method provided in the embodiment of the present invention uses a machine learning algorithm based on big data. Compared with the traditional credit enhancement model, it has a lower threshold, higher efficiency, and more accurate calculation results of the credit enhancement limit. At the same time, it takes into account special situations such as credit-free households and credit villages, and innovates the credit limit calculation method and credit enhancement method for new agricultural business entities, thereby improving the accuracy and reliability of the target credit limit.

[0067] Based on the above embodiment, the credit limit prediction model further includes a vertical federated learning entity alignment module; The vertical federated learning entity alignment module is used to perform vertical federated learning entity alignment on the data sets related to each new agricultural business entity, and then input them into the credit limit prediction model of the new agricultural business entity based on privacy computing.

[0068] Specifically, the credit limit prediction model also includes a vertical federated learning entity alignment module. The vertical federated learning entity alignment module uses the data from the three participating parties, namely the functional department management platform, the new agricultural business entity information management platform, and the bank, to perform anonymous intersection through the unique identifier of the new agricultural business entity, and find all the characteristic data of the same new agricultural business entity among the three participants. The data are all calculated locally and do not leave the domain.

[0069] Based on the above embodiment, the steps of acquiring the functional department characteristic data, the new agricultural business entity information management platform characteristic data and the bank characteristic data include: Step 301, performing data preprocessing and feature engineering on the first relevant data to obtain the functional department feature data; Step 302, performing data preprocessing and feature engineering on the second relevant data to obtain feature data of the new agricultural business entity information management platform; Step 303, performing data preprocessing and feature engineering on the third related data to obtain the bank feature data; The data preprocessing includes at least two of data cleaning, removing outliers, removing duplicate values, removing erroneous values, filling missing values, standardization and normalization; The feature engineering includes at least two of feature extraction, feature dimensionality reduction and feature balancing.

[0070] Specifically, the first relevant data can be subjected to data preprocessing and feature engineering to obtain functional department characteristic data, the second relevant data can be subjected to data preprocessing and feature engineering to obtain new agricultural business entity information management platform characteristic data, and the third relevant data can be subjected to data preprocessing and feature engineering to obtain bank characteristic data.

[0071] Data preprocessing includes at least two of data cleaning, removing outliers, removing duplicate values, removing erroneous values, filling missing values, standardization and normalization.

[0072] Feature engineering includes at least two of feature extraction, feature dimension reduction and feature balancing.

[0073] Here, the data sets related to each new agricultural business entity are first preliminarily screened for feature data for credit limit prediction through data preprocessing and feature engineering locally.

[0074] Based on any of the above embodiments, Figure 2 This is the second flow chart of the new agricultural business entity credit enhancement method provided by the present invention, such as Figure 2 As shown, the method includes: The first step is to preliminarily build a global model for predicting defaults of new agricultural operators.

[0075] The second step is to train a local model for predicting defaults of new agricultural operators.

[0076] The third step is to aggregate the default prediction models of new agricultural business entities.

[0077] The fourth step is to share the aggregated new agricultural business entity default prediction model with all participants and save it locally.

[0078] The fifth step is to determine whether the prediction accuracy meets expectations. If the prediction accuracy meets expectations, the default probability P of the new agricultural operator and the important values ​​of each feature are obtained through the default prediction model.β , otherwise, return to training the local model for predicting default of new agricultural business entities.

[0079] In the sixth step, each participant normalizes the feature data locally to obtain Z.

[0080] In the seventh step, each participant calculates the credit enhancement limits V1, V2, and V3 locally.

[0081] The eighth step is to obtain the target credit limit GV of new agricultural business entities through aggregation calculation.

[0082] Based on any of the above embodiments, Figure 3 This is a flow chart of the credit enhancement process for the new agricultural business entity provided by the present invention. Figure 3 As shown, the method includes: The first step is to gather relevant data from the functional department management platform, the new agricultural business entity information management platform, and the bank.

[0083] The second step is local data preprocessing and feature extraction.

[0084] In the third step, the privacy calculation model of the credit enhancement limit of the new agricultural business entity calculates the credit enhancement limit.

[0085] The fourth step is for banks to update the target credit limits for new agricultural operators.

[0086] Based on any of the above embodiments, the following is a specific embodiment: The credit enhancement base a set by the bank for a new agricultural operator in a certain scale planting area is RMB 300,000, and the credit enhancement limit is V top = 300,000 yuan, the basic unit of credit enhancement is 1,000 yuan, that is, u=1000. The original credit line of a new agricultural business entity A in the bank is 200,000 yuan, and the area where A is located is a credit village. Yuan, other information of A is as follows:

[0087] The default prediction model calculated the default rate to be 7.42%, and the important values ​​of each feature were calculated as follows:

[0088] According to the formula , the credit enhancement amount V1 calculated by the functional department management platform data = 60179

[0089] According to the formula , the credit enhancement amount V2 calculated by the data of the new agricultural business entity information management platform is 129870

[0090] According to the formula , the credit enhancement limit calculated by the bank's local data is V3=0

[0091] Credit enhancement line =

[0092] = =236000

[0093] New credit line = =5660000

[0094] Through credit enhancement, the credit line of new agricultural operator A in the bank increased from 200,000 yuan to 566,000 yuan.

[0095] The following is a description of the new agricultural business entity credit enhancement device provided by the present invention. The new agricultural business entity credit enhancement device described below and the new agricultural business entity credit enhancement method described above can be referenced to each other.

[0096] Based on any of the above embodiments, the present invention provides a new type of credit enhancement device for agricultural business entities. Figure 4 It is a structural schematic diagram of the new agricultural business entity credit enhancement device provided by the present invention, such as Figure 4 As shown, the device comprises: The acquisition unit 410 is used for multiple participants to respectively acquire data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from the functional department, second relevant data from the new agricultural business entity information management platform, and third relevant data from the bank; the data sets are vertically aligned and associated with the unique identifier of the new agricultural business entity; An input unit 420 is used to input the data set related to the new agricultural business entity into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit line enhancement line; A determination unit 430 is used to determine a target credit limit of the new agricultural business entity based on the credit enhancement limit; Among them, the default prediction model in the credit limit prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers. The local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and the subject label default rate and whether the data corresponds to the default label. The ratio of the training data set to the test data set is 7:3.

[0097] In the device provided by the embodiment of the present invention, multiple participants respectively obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from the functional department, second relevant data from the new agricultural business entity information management platform, and third relevant data from the bank; the data sets are vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; the data sets related to the new agricultural business entity are then input into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit line increase; finally, based on the credit line increase, the target credit line of the new agricultural business entity is determined. This process improves the information transparency between the bank and the new agricultural business entity through real-time convergence and fusion calculation of multi-channel information, and uses multi-source data to improve the credit line of the new agricultural business entity. The method is efficient, low-cost, objective and accurate, and the method is sustainable.

[0098] Based on any of the above embodiments, the credit limit prediction model includes the default prediction model and a credit enhancement limit calculation module for full-process multi-party collaboration; The default prediction model is used to: Inputting the data sets related to each new agricultural business entity into the default prediction model, obtaining the predicted value of the default rate of the new agricultural business entity and the importance of each feature within the default prediction model, and dividing the importance of each feature into a functional department data feature importance set, a new agricultural business entity information management platform data feature importance set and a bank data feature importance set according to different sources; The credit enhancement calculation module is used to determine a first credit enhancement based on the predicted default rate value and the functional department characteristic data and the characteristic importance corresponding to the functional department characteristic data, determine a second credit enhancement based on the predicted default rate value, the new agricultural business entity information management platform characteristic data and the characteristic importance corresponding to the new agricultural business entity information management platform characteristic data, and determine a third credit enhancement based on the predicted default rate value, the bank characteristic data and the characteristic importance corresponding to the bank characteristic data, and determine the credit enhancement based on the first credit enhancement, the second credit enhancement and the third credit enhancement; The functional department characteristic data are the characteristic data of new agricultural business entities from the functional departments; The new agricultural business entity information management platform characteristic data is the new agricultural business entity characteristic data from the new agricultural business entity information management platform; The bank characteristic data are characteristic data of new agricultural business entities from banks.

[0099] Based on any of the above embodiments, determining the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit, and the third credit enhancement limit includes: Determine the upper limit of credit enhancement that banks set for specific batches of new agricultural business entities, as well as the additional credit enhancement limit that banks grant to specific credit groups; Determining a non-default rate based on the predicted default rate value; The credit enhancement limit is determined based on the first credit enhancement limit, the second credit enhancement limit, the third credit enhancement limit, the credit enhancement upper limit, the additional credit enhancement limit, the non-default rate, and the credit enhancement base.

[0100] Based on any of the above embodiments, the credit limit prediction model further includes a vertical federated learning entity alignment module; The vertical federated learning entity alignment module is used to perform vertical federated learning entity alignment on the data sets related to each new agricultural business entity, and then input them into the credit limit prediction model of the new agricultural business entity based on privacy computing.

[0101] Based on any of the above embodiments, it further includes a characteristic data acquisition unit, wherein the characteristic data acquisition unit is specifically used to: Performing data preprocessing and feature engineering on the first relevant data to obtain the functional department feature data; Performing data preprocessing and feature engineering on the second relevant data to obtain feature data of the new agricultural business entity information management platform; Performing data preprocessing and feature engineering on the third related data to obtain the bank feature data; The data preprocessing includes at least two of data cleaning, removing outliers, removing duplicate values, removing erroneous values, filling missing values, standardization and normalization; The feature engineering includes at least two of feature extraction, feature dimensionality reduction and feature balancing.

[0102] Based on any of the foregoing embodiments, the determining unit 430 is specifically configured to: The target credit limit is determined based on the original credit limit and the credit enhancement limit.

[0103] Figure 5 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 5As shown, the electronic device may include: a processor 510, a communications interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communications interface 520 and the memory 530 communicate with each other via the communications bus 540. The processor 510 may call the logic instructions in the memory 530 to execute a credit enhancement method for a new agricultural business entity, the method comprising: a plurality of participants respectively obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from a functional department, second relevant data from an information management platform for new agricultural business entities and third relevant data from a bank; the data sets are vertically aligned and associated with entities using a unique identifier of the new agricultural business entity; the data sets related to the new agricultural business entity are input into a credit limit prediction model for the new agricultural business entity based on privacy computing to obtain a credit enhancement limit; based on the The credit enhancement line determines the target credit line of the new agricultural business entity; wherein, the default prediction model in the credit line prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers, and local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data is in default, and the ratio of the training data set to the test data set is 7:3.

[0104] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0105] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the new agricultural business entity credit enhancement method provided by the above methods, the method comprising: multiple participants respectively obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from new agricultural business entity information management platform and third relevant data from banks; the data sets are vertically aligned with the unique identifier of the new agricultural business entity; the data sets related to the new agricultural business entity are input into a privacy-based The credit line prediction model of the new agricultural business entity is calculated to obtain the credit enhancement line; based on the credit enhancement line, the target credit line of the new agricultural business entity is determined; wherein, the default prediction model in the credit line prediction model is obtained after local training based on local training data of each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers, and local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data is in default, and the ratio of the training data set to the test data set is 7:3.

[0106] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the credit enhancement method for new agricultural business entities provided by the above-mentioned methods, the method comprising: multiple participants respectively obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from functional departments, second relevant data from new agricultural business entity information management platforms, and third relevant data from banks; the data sets are vertically aligned and associated with the unique identifier of the new agricultural business entity; the data sets related to the new agricultural business entity are input into the credit limit of the new agricultural business entity based on privacy computing. A credit prediction model is used to obtain a credit enhancement line; based on the credit enhancement line, a target credit line of the new agricultural business entity is determined; wherein, the default prediction model in the credit line prediction model is obtained after local training based on local training data by each participating server, and then global coordination and aggregation are performed using the training parameters of all participating servers, and local training and global coordination aggregation are repeated many times; the default prediction model is a decentralized federated learning model; the training data set and test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data is in default, and the ratio of the training data set to the test data set is 7:3.

[0107] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0108] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A new type of credit enhancement method for agricultural business entities, characterized in that: include: The multiple participants obtain data sets related to each new agricultural business entity from their respective data sources; the data sets include first relevant data from the functional department, second relevant data from the new agricultural business entity information management platform, and third relevant data from the bank; The data set is vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; Inputting the data set related to the new agricultural business entity into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit enhancement line; Determine the target credit limit of the new agricultural business entity based on the credit enhancement limit; The default prediction model in the credit limit prediction model is obtained after each participant server performs local training based on local training data, and then uses the training parameters of all participant servers to perform global coordination and aggregation, and the local training and global coordination and aggregation are repeated many times. The default prediction model is a decentralized federated learning model; the training data set and the test data set of the local training data are composed of data related to each new agricultural business entity and labels of whether the entity corresponding to the data has defaulted, and the ratio of the training data set to the test data set is 7:

3.

2. The new agricultural business entity credit enhancement method according to claim 1 is characterized in that: The credit limit prediction model includes the default prediction model and a credit enhancement limit calculation module for full-process multi-party collaboration; The default prediction model is used to: Inputting the data sets related to each new agricultural business entity into the default prediction model, obtaining the predicted value of the default rate of the new agricultural business entity and the importance of each feature within the default prediction model, and dividing the importance of each feature into a functional department data feature importance set, a new agricultural business entity information management platform data feature importance set and a bank data feature importance set according to different sources; The credit enhancement limit calculation module is used to: Determine a first credit enhancement limit based on the predicted default rate value and the functional department characteristic data and the characteristic importance corresponding to the functional department characteristic data; determine a second credit enhancement limit based on the predicted default rate value, the new agricultural business entity information management platform characteristic data and the characteristic importance corresponding to the new agricultural business entity information management platform characteristic data; and determine a third credit enhancement limit based on the predicted default rate value, the bank characteristic data and the characteristic importance corresponding to the bank characteristic data; and determine the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit; The functional department characteristic data are the characteristic data of new agricultural business entities from the functional departments; The new agricultural business entity information management platform characteristic data is the new agricultural business entity characteristic data from the new agricultural business entity information management platform; The bank characteristic data are characteristic data of new agricultural business entities from banks.

3. The new agricultural business entity credit enhancement method according to claim 2 is characterized in that: The determining the credit enhancement limit based on the first credit enhancement limit, the second credit enhancement limit and the third credit enhancement limit comprises: Determine the credit enhancement base and credit enhancement ceiling set by the bank for a specific batch of new agricultural business entities, as well as the additional credit enhancement quota given by the bank to a specific credit group; Determining a non-default rate based on the predicted default rate value; The credit enhancement limit is determined based on the first credit enhancement limit, the second credit enhancement limit, the third credit enhancement limit, the credit enhancement upper limit, the additional credit enhancement limit, the non-default rate, and the credit enhancement base.

4. The new agricultural business entity credit enhancement method according to claim 2 is characterized in that: The credit limit prediction model also includes a vertical federated learning entity alignment module; The vertical federated learning entity alignment module is used to perform vertical federated learning entity alignment on the data sets related to each new agricultural business entity, and then input them into the credit limit prediction model of the new agricultural business entity based on privacy computing.

5. The new agricultural business entity credit enhancement method according to claim 2 is characterized in that: The steps of acquiring the functional department characteristic data, the new agricultural business entity information management platform characteristic data and the bank characteristic data include: Performing data preprocessing and feature engineering on the first relevant data to obtain the functional department feature data; Performing data preprocessing and feature engineering on the second relevant data to obtain feature data of the new agricultural business entity information management platform; Performing data preprocessing and feature engineering on the third related data to obtain the bank feature data; The data preprocessing includes at least two of data cleaning, removing outliers, removing duplicate values, removing erroneous values, filling missing values, standardization and normalization; The feature engineering includes at least two of feature extraction, feature dimension reduction and feature balancing.

6. The new agricultural business entity credit enhancement method according to any one of claims 1 to 5, characterized in that: Determining the target credit limit of the new agricultural business entity based on the credit enhancement limit includes: The target credit limit is determined based on the original credit limit and the credit enhancement limit.

7. A new type of credit enhancement device for agricultural business entities, characterized in that: include: An acquisition unit, used for multiple participants to respectively acquire data sets related to each new agricultural business entity from respective data sources; the data sets include first relevant data from functional departments, second relevant data from the new agricultural business entity information management platform, and third relevant data from banks; The data set is vertically aligned and associated with entities using the unique identifier of the new agricultural business entity; An input unit, used to input the data set related to the new agricultural business entity into the credit line prediction model of the new agricultural business entity based on privacy computing to obtain the credit line of the credit enhancement; A determination unit, configured to determine a target credit limit of the new agricultural business entity based on the credit enhancement limit; The default prediction model in the credit limit prediction model is obtained after each participant server performs local training based on local training data, and then uses the training parameters of all participant servers to perform global coordination and aggregation, and the local training and global coordination and aggregation are repeated many times. The default prediction model is a decentralized federated learning model; the training data set and the test data set of the local training data are composed of the data related to each new agricultural business entity and the labels of the subject labels corresponding to the data, the default rate and whether there is a default, and the ratio of the training data set to the test data set is 7:

3.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the new agricultural business entity credit enhancement method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the credit enhancement method for a new agricultural business entity as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, it implements the credit enhancement method for a new agricultural business entity as described in any one of claims 1 to 6.

Images