한국의 교통사고 발생행태 추정을 위한 교통사고 상관모형의 구축

Abstract

Motor vehicles have been continuously increasing with the development of the technologies and the increase of the travel demands, which has become a cause to increase the number of the traffic accidents and worsen the social problems from a social and economic point of view all over the world, especially in the developing countries. However, the number of traffic accidents in Korea as well as in the United States has continuously shown a trend to decrease since 2000, despite the increase of the motor vehicles registered. And a proper relationship has never been seen between the accident factor characteristics found in Korea.

The purpose in this study is to investigate 5-year traffic accident data in Korea from 2000 to 2004, analyze the accident factor characteristics based on those 5-year traffic accident data, construct the accident correlation models which could predict the accident occurrence patterns in the future based on the accident factor characteristics, and verify the accident correlation models constructed with the t-test and correlation analysis such that the accident occurrence patterns could be predicted by the proper relationship between traffic accident factor characteristics in Korea.

From the results of the development and validation of the accident correlation model with the accident factor characteristics, the conclusions were drawn as follows

ⅰ) Traffic accident patterns showed a uniform pattern with the 3-dimensional polynomial shift curve and a high determination coefficient (R2) for each accident factor characteristics. However, the percent of traffic accidents showed almost an equal percent for each year, even if traffic accidents have continuously reduced since 2000.

ⅱ) More traffic accidents occurred during the 2-hour period right after the rush hour in the afternoon than any other time periods, the daytime period than the nighttime period, the day before weekend than any other weekdays, and from March to October than any months, according to the periodic factor characteristics.

ⅲ) More traffic accidents occurred in the Capital Region than any other regions, in the intersection than any other locations, on the national highway than any other road types, in the suburban area than any other area types, and by passenger cars than any other vehicle types, according to the non-periodic factor characteristics.

ⅳ) Accident correlation models showed a very high explanatory power between the last 2 years before and after, and also a very high correlation between the expected and observed data, except for weekday, monthly, seasonal, area type factor characteristics based on the development and verification of the accident correlation models.

ⅴ) Traffic accident patterns showed a very high correlation with the 3-dimensional polynomial shift curves except for the monthly factor (R2=0.5231) in the periodic factor characteristics and the regional factor (R2=0.8542) in the non-periodic factor characteristics.

ⅵ) Traffic accident patterns particularly showed a much higher correlation with the 3-dimensional polynomial shift curves, when compared with the traffic accident patterns of the accident factor characteristics classified into the daytime and nighttime periods than those characteristics not classified into the daytime and nighttime periods.

Therefore, since the accident occurrence patterns in Korea were thought to show the equal patterns after 2004, especially except for the area type factor characteristics, the appropriate counter-measures to be established for the reduction of traffic accidents. Also, more accident occurrence patterns in the future were needed to be studied for the accident factor characteristics such as age, sex, or drinking of driver, if possible.