Apply Data Mining to the Rainfall Forecast for the Slope Land of Huafan University

• Main Authors: Chen-Cheng Lin, 林辰諶
• Other Authors: Zne-Jung Lee
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/76739593554265804398

碩士 === 華梵大學 === 資訊管理學系碩士班 === 103 === Differences in geography, climate, geology, and various forms of pollution cannot be avoided all kinds of natural disasters around the world, such as earthquakes, typhoons or tsunamis. The impact of natural disasters from landslides, mudslides, floods or slippery results in the loss of agricultural productions, life and property and so on. According to new research from global risk analytics company, Verisk Maplecroft, around the world in 2011 said that the United States, Japan, China and Taiwan were listed as extremely dangerous countries which suffered from natural disasters. Many papers concern landslides and rainfall threshold of landslides, but less discussions for threshold of inclinometer displacement. In this thesis, the data mining technology is applied to study how to establish the rules for observed rainfall and other relevant information in Huafan University. First, the regression analysis is used to predict rainfall. The results were obtained from the multiple regression formula and the prediction of rainfall. Then, classification and regression trees analysis were used to obtain rules. Obtained seven rules, contained several important variables, such as rainfall, insolation rate, temperature, X7, X18, and X36 tilt tube. The output of this approach can provide information for understanding the change of rainfall and can generate decision rules of rainfall threshold. Finally, it can predict rainfall patterns through time-series. These results could provide useful information to maintain the security for Dalun Mountain area.