China Safety Science Journal ›› 2018, Vol. 28 ›› Issue (8): 1-6.doi: 10.16265/j.cnki.issn1003-3033.2018.08.001

• Basic Disciplines of Safety Science and Technology •     Next Articles

Reduction and logistic regression model for prediction of rainfall landslides disaster

WEI Xingjun1, ZHAO Xiaomeng2, MA Changling1, LEI Xiangjie2   

  1. 1 School of Mechanical and Electrical Engineering,Shaanxi Energy Institute,Xianyang Shaanxi 712000,China;
    2 Shaanxi Provincial Climate Center,Xi' an Shaanxi 710014,China
  • Received:2018-05-10 Revised:2018-07-15 Online:2018-08-28 Published:2020-11-25

Abstract: In order to realize the rainfall-induced landslide disaster warning in Qinba mountain area,data on rainfall at different time points of landslide disasters from the nearest regional stations were collected and processed.The data set on rainfall characteristic landslide disaster was formed.A rainfall landslide knowledge expression system was defined.Knowledge reduction and prediction results feedback were used to filter feature attributes of rainfall.And then inducing factors of landslide disasters were determined.The logistic regression method was improved,the model generalization ability was enhanced and a reduced logistic regression model was established and compared with the uncertain naive Bayes model and the genetic neural network model.Experimental result shows that the reduced logistic regression model can effectively deal with the high-dimensional rainfall data and has a higher prediction precision accuracy.

Key words: rainfall, landslide disaster, knowledge reduction, logistic regression, P-R curve

CLC Number: