Research On Evidence Weight Algorithm Based On Logistic Regression

The maturity of spatial information technology has led to the rapid development of technologies related to spatial data mining.The evidence weight model based on uncertainty reasoning has an important position in the field of spatial data mining.In the actual application process,due to the restriction of the independence of evidence factors and the constraints of factor preprocessing,the accuracy of the evaluation results of traditional evidence weight model will be affected to a certain extent.This paper aims at the problems and deficiencies of the evidence weight model in spatial data mining,combined with the spatial data processing in the work process of data mining,based on the different spatial data mining model and application instances of in-depth analysis and research,the main steps of the existing evidence model was improved,so as to improve the accuracy of the model evaluation results.In order to solve the problem that primary evidence factors may not be related to the research objectives,the concept of correlation coefficient is introduced,and a method based on correlation coefficient to select evidence factors and determine the optimal critical value is proposed,which can effectively reduce the influence of factors irrelevant to the evaluation objectives on the final results;Aiming at the problem that the dimension and meaning of different evidence factor attribute data are different,the method of fuzzy conversion in fuzzy weight of evidence is used for reference.The noise removal and numerical normalization algorithm of evidence layer based on fuzzy membership degree is designed and realized;To solve the problem that the evidence factors do not satisfy the prerequisite of the evidence weight method,an improved algorithm for calculating the weight of evidence is proposed by combining the logistic regression model with the evidence weight model,and the logistic regression coefficient is used to correct the weight of evidence,which effectively weakens the influence of the independence on the evaluation results;Finally,a farmland production potential evaluation system based on the improved evidence right evaluation model is designed and realized,and an application example of farmland production potential evaluation is constructed by taking the land use situation of Weihai City as the research object.