Multi-dimensional risk feature strategy extraction method

Abstract

The invention relates to the technical field of data processing, in particular to a method for extracting a multi-dimensional risk feature strategy, which comprises the following steps of: collecting source data, preprocessing the source data, generating risk feature data, performing model training by adopting a LightGbm decision tree algorithm, generating a plurality of tree structures, and extracting branch nodes and segmentation thresholds in each tree structure; based on the branch node and the segmentation threshold value of each tree, performing binning on the client cluster, calculating the total quantity of sample clients and the total quantity of risk sample clients of each binning, and comparing the proportion of the risk sample clients of each binning with a set risk threshold value; and if a sub-box of which the risk sample customer proportion is greater than or equal to the risk threshold exists, determining that the branch node and the segmentation threshold of the tree are valid, and storing the branch node and the segmentation threshold as a multi-dimensional risk feature strategy combination feature. The method can flexibly, automatically and effectively combine the features in batches to form a quantitative analysis result and realize combined strategy extraction.

Description

technical field [0001] The invention relates to the technical field of data processing, in particular to a method for extracting multi-dimensional risk characteristic strategies. Background technique [0002] In all walks of life, risk management and control is very necessary and important, especially in the financial field. Risk management and control mainly includes risk management and control before, during and after lending. In the formulation of risk strategies in three different stages, in addition to the traditional In addition to expert experience, there are also quantitative analysis methods. Existing quantitative analysis methods are usually one-dimensional risk characteristic strategy method. In terms of the advantages and disadvantages of each method, the traditional expert experience method is simple and easy to implement, but it needs to be based on sufficient risk control experience and years of accumulation, which can be difficult to explain in words, requir...

AI Technical Summary

Problems solved by technology

In terms of the advantages and disadvantages of each method, the traditional expert experience method is simple and easy to implement, but it needs to be based on sufficient risk control experience and years of accumulation, which can be difficult to explain in words, requires high personnel and high cost, and at the same time, due to the lack of quantitative data. The evidence is difficult to be convincing; while the single-dimensional risk characteristic strategy is a commonly used method in designation, but it has not fully exploited the effectiveness of data sources and characteristics, and has not maximized the benefits in terms of data source cost and data source efficiency

Images