Decision tree construction method

Abstract

The invention discloses a decision tree construction method. A rule set space is recursively divided by a rule space division based decision tree algorithm to construct a decision tree. The method comprises: constructing an evaluation function used for quantitatively calculating a storage space of the decision tree and / or the balance of the decision tree; determining to-be-divided dimensions and corresponding to-be-divided rule sets for a current node; performing pre-division of different division frequencies on the current node in different to-be-divided dimensions to calculate values of the evaluation function, corresponding to different division modes; comparing the values of the evaluation function to determine division frequencies for the to-be-divided dimensions; and dividing the current node in the to-be-divided dimensions according to the division frequencies. Compared with the prior art, the decision tree more favorable for time and space complexity can be constructed based on the method provided by the invention; and meanwhile, according to the method provided by the invention, the scale of the whole decision tree can be controlled.

Description

Technical field [0001] The present invention relates to the field of information technology, in particular to a method for constructing a decision tree. Background technique [0002] With the continuous development of information technology, the data flow in the data processing system is also increasing. Since data of different nature types often correspond to different processing methods (for example, data with high processing priority needs to be processed first), it is necessary to classify the data before processing the data. [0003] In the current technical environment, the more commonly used data classification method is the decision tree classification method based on spatial division. The classification method based on the decision tree algorithm can not only implement data classification quickly and efficiently, but also the algorithm itself is simple and easy to implement, and can be used in various technical fields such as network security, network virtualization, and ...

AI Technical Summary

Problems solved by technology

However, the HyperCuts algorithm itself uses a local optimization algorithm, which cannot control the size of the entire decision tree.

[0006] In some cases, the decision tree with too large scale will make the memory usage of the application decision tree too large, which will make the algorithm infeasible at all.

Images