基于加权算法的土壤肥力状况的分析与研究

Analysis and research of the soil fertility status based on weighted algorithm

C4.5决策树算法可以利用训练数据集中的信息熵来构建分类模型,并通过对分类模型的简单查找来完成未知样本的分类,DBSCAN算法可以发现任意形状的簇,但由于土壤肥力数据的时空性、不平衡性和不确定性,两者都不能综合分析土壤肥力状况。因此,本文提出改进的算法来进行土壤肥力状况的分析与评价。该算法首先利用层次分析法平衡土壤养分各属性之间的差异;然后将标准化数据与C4.5决策树算法和DBSCAN算法相结合,通过运行效率与准确率来进行合理分类;其次,对两种加权算法进行比较。结果表明,改进的C4.5算法的准确率、算法效率都高于加权DBSCAN算法;用该算法综合评价连续多年精准施肥后土壤肥力的变化情况,与实际土壤养分逐年趋于平和的现象一致。最后,用加权DBSCAN算法对连续多年精准施肥的土壤肥力情况做分析。研究显示,利用改进的算法来处理土壤肥力状况的问题在分类稳定性上具有明显的优势。

C4.5decision tree algorithm can use the information entropy which used for training data set to construct a classification model,and the unknown samples can be classified through simple search of the model.DBSCAN algorithm can discover the clusters of arbitrary shape.However,due to the temporality and spatiality,the imbalance and uncertainty of the soil fertility data,both algorithms can not comprehensively analyze the soil fertility status.Therefore,the improved algorithm for the analysis and evaluation of soil fertility status was proposed.Firstly,AHP was used to balance the differences between various properties of soil nutrient.Then the standardized data was combined with C4.5decision tree algorithm and DBSCAN algorithm for a reasonable classification by running efficiency and accuracy.Secondly,comparing the two weighting algorithms,the accuracy and efficiency of improved C4.5algorithm are higher than the weighted DBSCAN algorithm.Using this method to comprehensively evaluating the changes in soil fertility with precision fertilization for many years,it is consistent with the phenomenon of actual soil nutrients which tend to calm by years.Finally,weighted DBSCAN algorithm was used to analyze the status of soil fertility with precision fertilization for many years.Research shows that the improved algorithm has obvious advantages in the classification stability.