چكيده به لاتين
increasing growth of vehicles, causing increase in the number of road accidents. Accident data singly is not so important and should be investigated and analyzed in order to be useful. Since these data are becoming more bulky, importance of data mining for accident data is determined. This study has been conducted to find appreciate and desirable method for data mining with the aim of classifying collision type and using Driver's gender, time passed from certification, type of certification, presence of human factors involved in the accident, presence of the road fault, presence of vehicle safety equipment, land use around collision location, road geometry, weather condition factors in order to classifying collision type with machine learning methods in data mining.
in this study logistic, MLP, SVM, Decision Tree and KNN classification methods is applied to classify collision type. Results of this study show that KNN method has best result in order to classification of collision type among other method discussed in this study. KNN method can classifies 87.3% of the test data correctly with the 0.58 correspond Cohen's kappa coefficient.
