聚类、粗糙集与决策树的组合算法在地力评价中的应用

 【目的】地力评价方法大多数有一定的主观性，较少考虑土壤各属性间的依赖关系。论文旨在采用数据挖掘方法，寻求地力等级划分的新方法。【方法】结合农安县耕地调查数据，应用K-means聚类方法、Johnson粗糙集属性约简算法与C4.5决策树算法相结合的优化算法评价地力等级。【结果】使用K-means聚类方法，得到最佳学习样本数；使用粗糙集属性约简和决策树相结合的方法，去掉了冗余属性7个，决策树模型共有节点317个，其中叶节点个数为159个，生成规则159条，模型准确率为82.08%。与未聚类和未约简的方法相比，决策树结点个数减少41.62%。【结论】使用该组合算法，在保证模型准确率的同时，降低了算法的时间和空间复杂性，提高了挖掘效率。

Applied Research of Combinatorial Algorithm of Clustering, Rough Set and Decision Tree Method in Productivity Evaluation

【Objective】 Fertility evaluation method has a certain subjective and less considers the dependence relation among soil attributes. This paper is aimed to seek a new method of productivity evaluation by data mining method. 【Method】 Based on Nong’an cultivated land survey data, the paper used optimization algorithm of K-means clustering method, Johnson rough set attribute reduction algorithm and C4.5 decision tree algorithm to evaluate the productivity grade. 【Result】 The best learning samples are obtained by using K-means clustering method. Rough sets are used in soil attribute reduction, and 7 soil redundant attributes are removed. The decision tree model has 317 nodes and 159 leaf nodes, extracts 159 rules, model accuracy is 82.08%. The decision tree node number decreased by 41.62% compared with no-clustering and no-reduction approaches. 【Conclusion】 Using the combination algorithm, while the accuracy of the model is ensured, the algorithm time and space complexity are reduced and the mining efficiency is improved.