Application of Covariance and Correlativity Coefficient to the Designing of Decision Trees
Zhao Xiang · 2003
Information gain is the measurement of the attributes selection in classical decision tree algo-rithm-ID3, but the attributes with high information gain is not always the valuable attributes. In this paper, covariance and correlativity coefficients are used to describe the correlation of random variable, and a new decision tree construction algorithm is proposed.