Research on Rough Set and Decision Tree Method Application in Evaluation of Soil Fertility Level

Abstract

Clustering, rough sets and decision tree theory were applied to the evaluation of soil fertility levels, and provided new ideas and methods among the spatial data mining and knowledge discovery. In the experiment, the rough sets - decision tree evaluation model establish by 1400 study samples, the accuracy rate is 92% of the test. The results show :model has good generalization ability; using the clustering method can effectively extract the typical samples and reducing the training sample space; the use of rough sets attribute reduction, can remove redundant attributes, can reduce the size of decision tree decision-making model, reduce the decision-making rules and improving the decision-making accuracy, using the combination of rough set and decision tree decision-making method to infer the level of a large number of unknown samples.