APPLICATION OF MULTIPLE REGRESSION ANALYSIS AND NEURAL NETWORK TO THE PREDICTION OF ROAD FLOOD DEPTH

Abstract

To predict road flood depth helps the road user to choose a safe route and road administrators to efficiently manage roads. In this paper, the road water level in Saitama City and the rainfall from radar rain gauge are used as explanatory variables, and the future road water level is used as the objective variable. One of the prediction model uses multiple regression analysis and the other neural network. We evaluated the observation data for three years with these models. And the effectiveness of the prediction model constructed at the low water level for the high water level were evaluated using October 2019 including Typhoon No. 19 in 2019 as the test data.