An Arbitrary Gini Index for the Redundant Feature Datasets Analysis

Objectives: Knowledge Discovery methods get more accurate results when the dimensionality of the data is subsided; dimensionality is an important aspect of any data. Several algorithms have been proposed to increase the accuracy, but most of them generate complex models as the size of the data is extremely large. Objective of this paper is to build a simple model to get high accuracy. Method: In order to increase the accuracy of the Knowledge Discovery methods by substituting the dimensionality, we introduce a novel heuristic functionality, Arbitrary Gini Index (ArGI). Findings: We evaluated the performance of ArGI on the real world datasets. The experiment on the ten real world data sets analysis shows 60% data sets are more accurate for ArGI and 40% for Gini Index. Applications: It is expecting that the applications of ArGI will show a better approach in the real world learning tasks.