Method and device for automatically binning data

Abstract

The present invention relates to the technical field of data processing, and in particular to a method and device for automatically binning data. The method specifically includes: acquiring basic feature data and binning conditions input by users, and bringing the binning conditions into predefined functions to obtain the target Function, determine the initial vector according to the binning condition, and bring the initial vector into the objective function to determine the search direction for the basic feature data. Then, with the initial vector as the reference point, adjust the initial vector according to the search direction and bring it into the objective function to obtain the corresponding function value. When the difference between the latter function value and the current function value is less than the preset convergence accuracy, determine the latter function value The corresponding initial vector is used as the segmentation point, and finally the basic feature data input by the user is binned according to the determined multiple segmentation points. Through this solution, fast binning can be realized, so that the degree of correlation between each binning is the lowest, and it is convenient to perform an objective scoring operation on the data input by the user.

Description

technical field [0001] The present invention relates to the technical field of data processing, in particular to a method and device for automatically binning data. Background technique [0002] With the development and popularization of big data and artificial intelligence technology, more and more financial institutions have increased their emphasis on machine learning, gradually transforming traditional management methods based on manual decision-making into data-driven intelligent decision-making. Especially in banking personal financial business, such as credit card business, consumer financial business and other fields, due to the small amount of a single transaction, high application frequency, and high timeliness requirements, the traditional manual approval method cannot meet business demands. Using machine learning methods for risk management, especially the scorecard model based on logistic regression, is gradually being adopted by banks because of its easy-to-int...

AI Technical Summary

Problems solved by technology

[0005] The above two types of automatic binning methods are too sensitive to iteration termination conditions, such as tree depth, minimum bin capacity, etc., which may easily cause overfitting problems. At the same time, the two types of automatic binning methods have limited support for constraints (such as certain types of data must is a box, specify the box interval, etc.), which cannot fully meet the needs of the binning problem in the actual modeling process

Images