论文部分内容阅读
在充分利用土壤类型、土地利用方式、岩性类型、地形、道路、工业类型等影响土壤质量主要因素,准确获取区域土壤质量的空间分布特征的基础上,采用互信息理论对13个辅助变量(岩性类型、土地利用方式、土壤类型、到城镇的距离、到道路的距离、到工业用地的距离、到河流的距离、相对高程、坡度、坡向、平向曲率、纵向曲率和切线曲率)进行筛选,然后通过决策树See5.0预测研究区土壤质量.结果表明:影响研究区土壤质量的主要因素包括土壤类型、土地利用方式、岩性类型、到城镇的距离、到水域的距离、相对高程、到道路的距离和到工业用地的距离;以互信息理论选取的因子为预测变量的决策树模型精度明显优于以全部因子为预测变量的决策树模型,在前者的决策树模型中,无论是决策树还是决策规则,分类预测精度均达到80%以上.互信息理论结合决策树的方法在充分利用连续型和字符型数据的基础上,不仅精简了一般决策树算法的输入参数,而且能有效地预测和评价区域土壤质量等级.
On the basis of taking full advantage of the main factors influencing soil quality, such as soil type, land use pattern, lithology type, topography, road and industrial type, and accurately obtaining the spatial distribution characteristics of regional soil quality, we use the mutual information theory to analyze 13 auxiliary variables Lithology type, land use pattern, type of soil, distance to town, distance to road, distance to industrial land, distance to river, relative elevation, grade, aspect, flat curvature, longitudinal curvature and tangent curvature) The results showed that the main factors influencing the soil quality in the study area include soil type, land use type, lithology type, the distance to the town, the distance to the water area, the relative The distance to the road and the distance to the industrial land. The precision of the decision tree model with the factor selected as the predictor by the mutual information theory is obviously better than the decision tree with all factors as the predictor. In the former decision tree model, Whether it is decision tree or decision rule, the accuracy of classification prediction reaches more than 80% Tree-Based Methods in continuous and full use of the character data, not only the general streamlining of the decision tree algorithm input parameters, and can effectively predict and soil quality grade evaluation region.