Research On Influence Factors Of Weathering Based On Logistic And Cart Model

In recent years, along with the tourism sustainable development and global climate change, the dependent environment of Yungang grottoes has changed significantly. At the same time, because of complex of natural conditions and inadequate understanding of the mechanism of cultural relics weathering in Yungang grottoes, the protection of Yungang grottoes is always the problems of perplexs the heritage conservation work. So it is necessary to do research on the influencing factors of weathering on the rocky relics in the Yungang grottoes.To discussing the factors affecting the rocky cultural relics weathering, this research used the decision tree CART classification algorithm model of the data mining classification method and the unconditional Logistic regression model of the traditional regression analysis method. Comparing the results between the CART model and the Logistic model, I further discussed that the CART model and the Logistic model can complement each other. First, through investigating, expert consulting, literature researching, I learned about the weathering mechanisms, weathering processes and possible influencing factors of the rocky relics, and got the relevant data by various channels. Secondly, by the method of Unconditional Logistic regression analysis, I established the Unconditional Logistic regression model of weathering influence factors, and verified the results of the model. Again, under the condition of clearing the principle of CART algorithm of decision tree, I established the decision tree model with of the weathering influence factors by this algorithm, and verified the result. Finally, by using the AUC forecast probability of each model, I evaluated the accuracy of the model classification, and fitted new Logistic regression model using the information provided by the decision tree model, and compared the differences of the effect factors and the classification accuracy screened by the new Logistic model and the original one.In this paper, I applied the decision tree classification method of the data mining into the research of the stone relics weathering influential factors. Combined with the unconditional Logistic model, I can more fully utilize the relevant data, get more comprehensive results. This provides a new study idea for the stone relics weathering influence factors research, and has certain theoretical and practical reference value.