Multi-Level Indicators Reduction Approach for Complex Index System

Abstract

The reduction of the index system at home and abroad mainly starts from the basic statistics and soft computing theory, but these methods are not always effective for complex index system. In this paper, a new multi-level indicators reduction model based on complex index system has been proposed through in-depth analysis of the multi-level environment and growth factors which affecting crop growth. First, exception handling model was introduced to determine indicators abnormalities or not in this method, then make a preliminary reduction in indicators of index layer from the point of multi-level statistical; Second, through clustering reduction model, and a further perfection of the index system is discussed; Finally, the optimal reduction index system of crop growth evaluation is proposed. The experiment results show that it can get the optimal reduction index system of crop growth evaluation by this method, and the optimal index system can cover 90% of the index system.