Novel decision tree classification method based on J divergence

Abstract

The invention provides a novel decision tree classification method based on J divergence. The method comprises the following steps of S1, standardizing an input sample data set; S2, setting a divisiontermination condition of the sample data set or the data subset, i.e., generating a condition that the leaf node does not continue to be divided according to the state of the current node by the decision tree; S3, preparing to divide the original sample data set, and creating a root node of the decision tree; S4, dividing the sample data set, splitting nodes of the decision tree and generating adecision rule according to a division criterion; S5, continuously dividing the sample subset obtained by dividing each time according to the steps S2, S3 and S4 by adopting a recursive mode to obtaina new node; and S6, discriminating and predicting the category value of the sample X * with the unknown category label value. According to the method, J divergence information measurement is innovatively used for generation of a division criterion, so that the prediction accuracy of the decision tree classification algorithm is improved.

Description

technical field [0001] The present invention relates to the cross technical field of information theory and data mining, in particular to a novel decision tree classification method based on J-divergence. Background technique [0002] With the development and progress of technologies such as wireless mobile communications, the Internet, and various smart terminal devices, massive amounts of data are being generated and collected exponentially and explosively. How to discover and extract useful knowledge or rules from massive big data is a series of issues worth considering. These problems all involve a key technology in big data technology, that is, data mining technology. The classification problem is a typical problem to be solved by data mining, and it is widely used in practical application scenarios such as spam recognition, text and image recognition, financial risk control and credit card fraud, online advertisement placement, and recommendation systems. Most of the...

AI Technical Summary

Problems solved by technology

[0006] In view of the defects in the prior art, the purpose of the present invention is to provide a novel decision tree classification method based on J-divergence, which is applied to the classification of data sample sets, so as to solve the problem of poor classification prediction accuracy of the existing decision tree classification method. good technical questions

Images