【摘要】 为了探寻影响水库水环境的主要因素,以河南省38座水库作为典型,根据其2012年的水质监测数据,采用CART决策树方法,以其中的28座为训练集,通过Modeler特征选择选出7个主要影响因素作为变量,水库水质为目标变量,建立模型,并用全部38座水库对模型进行了验证,结果表明:模型训练的准确率为100.0%,验证的准确率为94.74%;破碎度对模型的影响程度为0.43,单位面积农药施用量、单位面积施肥量、单位面积工业产值、人口密度、单位面积工业废水排放量等5个因素对模型的影响程度均为0.11,单位面积GDP对模型的影响最小,仅有0.02.决策树方法可以应用于水库的水质影响因素研究.更多还原

【Abstract】 Taking thirty-eight reservoirs of Henan Province for typical examples, and the classification and regression tree(CART) decision tree method is adopted, and then based on the water quality monitoring data of 2012, the main factors affecting reservoir water environment are explored. Seven main factors are chosen as variables by the module of feature selection from Modeler software to build up model with twenty-eight reservoirs as training set and water quality as the goal; the model is verified by whole reservoir numbers. The research results show that the training accuracy of the model is 100%, and the rate of verifying was 94.74%; the influence of fragmentation degree is the largest with 0.43; the influence degree of applying pesticide per unit area, fertilizer rate per unit area, industrial output per unit area, population density, industrial wastewater emissions per unit area on the model are all 0.11; the effects of GDP per unit area on the model is minimum, only 0.02. It is concluded that the decision tree method can be applied to study the influencing factors on water quality for reservoirs.更多还原