关键词: ID．3 量化 特征直方图 均衡化 平均深度

Abstract:

The ID. 3 algorithm is a classic decision tree algorithm, and the imbalance of the samples＇ distribution has a great influence on the tree＇s structu re and its classification effect. The paper uses the technology of feature histogram equalization to transfer the laws to the uniform distribution law i n [0,1], based on the analysis of the cell recognization feature database under the microscope. Experiments prove that the convergence speed of ID. 3＇s error rate is qulckened,and the average depth of the decision tree is shortened by this method.

Key words: ID. 3, quantization, feature histogram, equalization, average depth