Post-loan management committee case allocation method based on micro-service architecture

Abstract

The invention relates to the field of post-loan management committee case allocation, and discloses a micro-service architecture-based post-loan management committee case allocation method, which comprises the following steps of: dividing post-loan management committee case allocation into a plurality of micro-service modules; the micro-service module comprises a case information management module, a case commitment mechanism management module and a distribution strategy management module; the case information management module collects case information and extracts case features; the state information of the commitment institution is updated in real time in the commitment institution management module; a multi-dimensional case allocation strategy is adopted in the allocation strategy management module to perform case allocation based on an intelligent allocation algorithm; and monitoring a collection result in real time and performing model optimization. The method can combine the characteristics of the case and the characteristics of the committed case institution for distribution, maximizes the income, and improves the management efficiency.

Description

Technical Field

[0001] The present invention relates to the field of post-loan management case allocation, and in particular to a post-loan management case allocation method based on a microservice architecture. Background Art

[0002] With the rapid development of financial technology, the case handling needs of financial institutions in post-loan management are becoming increasingly complex. Traditional post-loan management systems usually adopt a centralized case allocation method, that is, all cases are handled by a single or a few commissioned case agencies (such as collection companies, law firms, etc.). However, this centralized allocation method has many problems, especially when faced with a large number of cases, it is easy to lead to uneven case distribution, which in turn affects collection efficiency and customer experience. In order to solve these problems, the present invention proposes a post-loan management commissioned case allocation method based on a microservice architecture, which aims to reasonably allocate cases through multiple dimensions, ensure that case allocation is more scientific and efficient, and avoid concentrating cases with large amounts in a single commissioned case agency. Summary of the invention

[0003] The present invention aims to provide a post-loan management case allocation method based on a microservice architecture, so as to allocate the case in combination with the characteristics of the case and the case-entrusting institution, so as to maximize the benefits and improve the management efficiency.

[0004] In order to achieve the above object, the present invention adopts the following technical scheme:

[0005] A post-loan management case allocation method based on microservice architecture, comprising:

[0006] S1, dividing the post-loan management case assignment into several micro-service modules; the micro-service modules include a case information management module, a case assignment agency management module, and an assignment strategy management module;

[0007] S2, collecting case information and extracting case features in the case information management module;

[0008] S3, updating the case entrusting agency status information in real time in the case entrusting agency management module;

[0009] S4, in the allocation strategy management module, a multi-dimensional case allocation strategy is used to allocate cases based on an intelligent allocation algorithm;

[0010] S5, monitors collection results in real time and optimizes the model.

[0011] The principles and advantages of this solution are: in actual application, in the existing post-loan management case allocation system, cases with large amounts of money are often concentrated in a few case-assigning agencies, resulting in excessive workloads for these agencies and affecting the collection effect; the capabilities and characteristics of different case-assigning agencies vary, and the traditional allocation method fails to fully utilize the advantages of each agency, resulting in a waste of resources; the number of post-loan management cases is huge, and the case allocation process is cumbersome, which cannot quickly respond to market demand, affecting the overall collection efficiency. In this application, the post-loan management case allocation is divided into several microservice modules, so that each service module is relatively independent, which improves the technical flexibility of each service module, and obtains case information and case-assigning agency information through the case information management module and the case-assigning agency management module respectively, and then uses a multi-dimensional case allocation strategy in the allocation strategy management module to allocate cases based on an intelligent allocation algorithm, which is convenient for allocation based on the characteristics of the case and the case-assigning agency, so as to maximize and rationalize the benefits and improve management efficiency.

[0012] Preferably, as an improvement, the case characteristics include basic information of the borrower, loan information, repayment record, and collection results.

[0013] Technical effect: The above typical features make it easier to accurately assign cases based on the cases.

[0014] Preferably, as an improvement, the commissioning agency status information includes the commissioning agency's collection capability, regional coverage, and workload dimension.

[0015] Technical effect: It is easy to accurately grasp the capability information of the commissioning agency.

[0016] Preferably, as an improvement, the intelligent allocation algorithm is an intelligent allocation algorithm based on machine learning, including:

[0017] S41, collect historical case data, and clean and pre-process the historical case data;

[0018] S42, extract key features from historical data;

[0019] S43, using machine learning algorithms to train historical data and build a case allocation model;

[0020] S44, based on the information of the current case and the status of the commissioning agency, calls the trained model in real time to predict the optimal allocation plan and allocates the case to the most appropriate commissioning agency.

[0021] Technical effect: By learning from historical case data, the allocation strategy is automatically optimized to ensure the rationality and fairness of case allocation.

[0022] Preferably, as an improvement, the machine learning algorithm is a random forest.

[0023] Technical effect: By integrating multiple decision trees, overfitting is reduced, training is parallelized, and the stability and accuracy of the model are improved.

[0024] Preferably, as an improvement, the S43 includes:

[0025] S431, randomly sample the data set using the bootstrap method to generate multiple different training sets;

[0026] S432, when splitting each node, only some randomly selected features are considered;

[0027] S433, constructing a decision tree for each training set generated by sampling;

[0028] S434, obtain the prediction category given by each decision tree, and the random forest selects the commissioning agency with the most votes as the prediction allocation result through a voting mechanism.

[0029] Technical effect: It is easy to increase the diversity of the model and reduce the correlation between trees, thereby improving the accuracy of the results.

[0030] Preferably, as an improvement, in S432, the splitting criterion is:

[0031]

[0032] Among them, D is the dataset of the current node, k is the number of categories, and p k is the sample proportion of the kth category.

[0033] Technical effect: The smaller Gi(D) is, the higher the purity of the node is. The splitting criteria make it easier to determine how to select the optimal features and splitting points.

[0034] Preferably, as an improvement, S5 includes:

[0035] S51, record the case collection results;

[0036] S52, statistically comparing the collection results with a preset threshold;

[0037] S53, when the collection result exceeds the threshold, the model is retrained using new data.

[0038] Technical effect: It is easy to optimize the model in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic structural diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The following is further described in detail through specific implementation methods:

[0041] The embodiment is basically as shown in the attached Figure 1 As shown:

[0042] A post-loan management case allocation method based on microservice architecture, comprising:

[0043] S1, divide the post-loan management case assignment into several microservice modules; each microservice is an independent, small module, each service is responsible for implementing specific business functions, and each service module can select a suitable database and technical framework, which improves the technical flexibility of each service module. The microservice modules include but are not limited to case information management module, case assignment agency management module, and allocation strategy management module; the case information management module is responsible for collecting and managing the basic information of post-loan cases, including case amount, borrower information, overdue days, repayment history, etc. The case assignment agency management module is responsible for maintaining the basic information of the case assignment agency, including agency name, contact information, collection capabilities, historical performance, regional coverage, etc. The allocation strategy management module is responsible for defining and managing case assignment strategies, supporting multi-dimensional allocation rules, and ensuring the rationality and fairness of case assignment.

[0044] S2, collect case information in the case information management module and extract case features; case features include borrower basic information (age, gender, occupation, income), loan information (loan amount, term, interest rate), repayment record (whether overdue, number of overdue days, repayment frequency) and collection results (whether successful recovery, amount recovered).

[0045] S3, in the case agency management module, the case agency status information is updated in real time; the case agency status information includes the collection ability, regional coverage, and workload dimension of the case agency. Based on the above case agency status information, it is convenient to reasonably allocate cases according to the characteristics of the case agency, ensuring that each agency can maximize its efficiency in the field it excels in. For example, for an agency that is good at handling large-value cases, high-value cases are allocated first; for an agency that is good at handling small-value cases, more small-value cases are allocated.

[0046] S4, in the allocation strategy management module, a multi-dimensional case allocation strategy is used to allocate cases based on an intelligent allocation algorithm; the intelligent allocation algorithm is an intelligent allocation algorithm based on machine learning. In this embodiment, the machine learning algorithm is preferably a random forest, including:

[0047] S41, collect historical case data, and clean and preprocess the historical case data; specifically, process missing values ​​using mean filling, median filling, mode filling or interpolation methods; detect and process outliers through box plots or Z-Score methods; remove duplicate data to ensure that there are no duplicate case records in the data set.

[0048] S42, extracting key features from historical data; in this embodiment, features related to collection effects are selected based on business experience and statistical analysis.

[0049] S43, use machine learning algorithms to train historical data and build a case allocation model; S43 specifically includes: S431, use the bootstrap method to randomly sample the data set to generate multiple different training sets: extract some samples from the training set with replacement as training data for the decision tree, assuming that the original training set has (N) samples, and randomly extract (N) samples each time, some samples will be selected multiple times, while other samples will not be selected. S432, when splitting at each node, only consider some randomly selected features; assuming that there are a total of (p) features, randomly select each time when splitting The dataset is recursively split into subsets based on the selected features and splitting criteria until the stopping condition is met. The splitting criteria are:

[0050]

[0051] Among them, D is the dataset of the current node, k is the number of categories, and p k is the sample proportion of the kth category.

[0052] S433, construct a decision tree for each training set generated by sampling; repeat to construct multiple decision trees to form a random forest. S434, obtain the prediction category given by each decision tree, and the random forest selects the commissioning agency with the most votes as the prediction allocation result through a voting mechanism.

[0053] S44, based on the information of the current case and the status of the commissioning agency, calls the trained model in real time to predict the optimal allocation plan and allocate the case to the most appropriate commissioning agency. By allocating cases rationally in multiple dimensions, the fairness and efficiency of case allocation are ensured, and cases with large amounts are avoided from being concentrated in a single commissioning agency, thereby improving the overall effect of post-loan management.

[0054] As new data continues to accumulate, the allocation model needs to be updated in a timely manner to ensure that the allocation model is in a relatively optimal state. Therefore, S5 is also included to monitor the collection results in real time and optimize the model. Specifically, S5 includes: S51, recording the collection results of the case, including whether the collection is successful and the amount of collection; S52, statistically comparing the collection results with the preset threshold; the preferred statistical values ​​in this example include the frequency of the collection results exceeding the preset threshold and the total number of collections; S53, when the statistical value of any collection result exceeds the threshold, the model is retrained using new data.

[0055] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and / or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several modifications and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A post-loan management case allocation method based on microservice architecture, characterized in that ,include: S1, dividing the post-loan management case assignment into several micro-service modules; the micro-service modules include a case information management module, a case assignment agency management module, and an assignment strategy management module; S2, collecting case information and extracting case features in the case information management module; S3, updating the case entrusting agency status information in real time in the case entrusting agency management module; S4, in the allocation strategy management module, a multi-dimensional case allocation strategy is used to allocate cases based on an intelligent allocation algorithm; S5, monitors collection results in real time and optimizes the model.

2. According to the method for allocating post-loan management cases based on microservice architecture according to claim 1, it is characterized by: The case characteristics include borrower basic information, loan information, repayment record, and collection results.

3. According to the method for allocating post-loan management cases based on microservice architecture as claimed in claim 1, it is characterized by: The commissioning agency status information includes the commissioning agency's collection capabilities, regional coverage, and workload dimensions.

4. According to the method for allocating post-loan management cases based on microservice architecture as claimed in claim 1, it is characterized in that: The intelligent allocation algorithm is an intelligent allocation algorithm based on machine learning, including: S41, collect historical case data, and clean and pre-process the historical case data; S42, extract key features from historical data; S43, using machine learning algorithms to train historical data and build a case allocation model; S44, based on the information of the current case and the status of the commissioning agency, calls the trained model in real time to predict the optimal allocation plan and allocates the case to the most appropriate commissioning agency.

5. According to the method for allocating post-loan management cases based on microservice architecture as claimed in claim 4, it is characterized by: The machine learning algorithm is random forest.

6. A post-loan management case allocation method based on microservice architecture according to claim 5, characterized in that: The S43 includes: S431, randomly sample the data set using the bootstrap method to generate multiple different training sets; S432, when splitting each node, only some randomly selected features are considered; S433, constructing a decision tree for each training set generated by sampling; S434, obtain the prediction category given by each decision tree, and the random forest selects the commissioning agency with the most votes as the prediction allocation result through a voting mechanism.

7. A post-loan management case allocation method based on microservice architecture according to claim 6, characterized in that: In S432, the splitting criteria are: Among them, D is the dataset of the current node, k is the number of categories, and p k is the sample proportion of the kth category.

8. The method for allocating post-loan management cases based on microservice architecture according to claim 1 is characterized by: The S5 includes: S51, record the case collection results; S52, statistically comparing the collection results with a preset threshold; S53, when the collection result exceeds the threshold, the model is retrained using new data.

Images