Improvement of SLIQ Algorithm and its Application in Evaluation

Abstract

In view of the high operation cost of Gini index for SLIQ algorithm while searching the optimal split scheme, and from the objective distribution of data, the thesis introduces the concept of Density of Data Distribution. The new SLIQ algorithm, improved on the base of Gini index, is adopted in synthetic evaluation. The results in the example demonstrate that the number of the Gini index is reduced greatly after the improvement of SLIQ algorithm while searching the optimal split scheme. The cost of sorting and the optimal split point is also cut down. So the size of decision tree is simplified. Thus the evaluating objectives are classified, and the evaluating results are arranged in order and then the ordering and classified synthetic evaluation is realized.