Principal component analysis and Fisher discriminant analysis of environmental and ecological quality, and the impacts of coal mining in an environmentally sensitive area

Most discriminant methods do not consider the problem of misjudgment related to the superposition of information from different discriminant indexes. Therefore, we used principal component and Fisher discriminant analysis to model, assess, and classify environmental and ecological quality, and the impacts of coal mining. The analysis uses the following input parameters as discrimination indexes: geomorphology, water depth, thickness of the phreatic water layer, bedrock thickness above the uppermost coal seam, and thickness of the uppermost coal seam. Twenty-three datasets from the Yushenfu coal mine area, Shaanxi Province, China, were used to train the model. The validity of the model was tested by the backward substitution method, and the misjudgment rate was zero. Seven datasets were then used as test samples in a support vector machine model. Our results show that it is feasible to predict the environmental and ecological impacts of coal mining with principal component analysis and Fisher discriminant analysis, which can effectively eliminate the interaction between the sample variables. This results in a more accurate assessment of mine environmental quality and represents a new method for predicting the impacts of coal mining in environmentally sensitive areas.