Abstract: Aiming at the problem that the algorithm for traditional C45 decision tree classification algorithm needed to be scanned several times, resulting in defects of running low efficiency, the author proposed a new improved C45 decision tree classification algorithm by optimizing the logarithmic operation related information gain derivation algorithm in order to reduce the running time of the decision tree classification algorithm. And the simple split attribute of the continuous attributes in the traditional algorithm was improved to the optimal partition point splitting processing in order to improve the efficiency of
the algorithm. Experimental results show that compared with the traditional C45 decision tree classification algorithm, the improved C45 decision tree classification algorithm greatly improves the execution efficiency and reduces the demand space.

Key words: C45 decision tree, data mining, discriminative ability measure, continuous attribute, classification algorithm