The ship accidents were caused by various factors. Various factors consist of human factors and natural factors. In this paper, we focus on natural factors. The purpose of this paper is to analyze ship accidents in Busan and Incheon and analyze the effects of natural factors on ship accidents using logistic regression analysis and decision trees. The Korean Peninsula is surrounded by three different seas, and each sea has different natural factors. The data of natural factors were obtained from the Korean Maritime Safety Tribunal and the Korea Meteorological Administration. Each data of tow factors was combined. Ship accident data, general weather data, and marine weather data were refined by propensity score matching(PSM) method. We observed that significant digging, wind speed, daily precipitation, and barometric pressure affected ship accidents in Busan. Also we found that wind speed, daily precipitation, and temperature affected ship accidents in Incheon. In future studies, the factors of ship accidents should be modeled by considering the regional difference.