Designing Accident Possibility Prediction Model for Accident Prevention

Although recorded accident data can used to examine similar cases and show caution about accident, various analysis, which is the nature of accident and is relationship between accident and surrounding environmental factors, have been insufficient. Therefore, accident data with the related factor such as weather information must be needed in order to provide an accident possibility per specific areas using risk prediction model. This paper aims to develop analysis model for deriving accident possibility per area using weather information and accident or its report data. Past weather information collected about nine year using AWS (Automatic Weather Station) of the Korea Meteorological Administration. Accident and its report data are collected by verifying the received report and exanimated accident in Korea Gas Safety Corporation for about nine years. Our analysis model of accident possibility is designed and developed by applying, K-NN(nearest neighbor) algorithm on basis of probability between accident and weather information about each of twenty three areas individually