Application of evaluation in farmland with decision tree model based on clustering

To choose reasonable learning samples and to enhance the validity of decision tree in farmland evaluation, a new evaluation method was proposed in the paper. The model of farmland evaluation was constructed by improved the algorithm of C5.0 decision tree based on clustering methods for choosing learning samples. Taking farmland in Longchuan County, Guangdong Province as example, six kinds of learning samples were chosen from clustering results by means of experimentation, and 4000 samples were finally learning samples. The evaluation model of decision tree was improved with cost weight, its finally prediction accuracy reached 94.92%, which was satisfied with practical demand. The results show that the integrated mode using clustering method and decision tree is feasible for farmland evaluation. The constructed evaluation model has high accuracy, robustness and comprehension.