Handwriting near miss report, which is a traditional way in the Japanese coastal tankers, inhibits the practical and quick analysis; as a result, it takes long time for effective feedback.
In this study, first, the authors developed the web system for reporting near miss to solve these problems. The issues emerged in the process of developing and operating the system were discussed from both technical and business point of view.
Secondly, the digitized data collected by the developed system were analyzed by data mining technique; near miss data can be categorized into 21 factors based on 4M(Man, Machine, Media and Management) as typical items. Additionally, it was clarified that reporting the hypothetical near miss, which a crew expected to likely occur, led to developing risk prediction skills.
Finally, the deviations obtained from report proportions between actual near miss and hypothetical one on the above 21 factors was studied; it was found that the deviations can be used as an index to extract items for the criterion of safety education.
