Prediction and Analysis of the Severity and Number of Suburban Accidents Using Logit Model, Factor Analysis and Machine Learning: A case study in a developing country

Abstract The purpose of this study is to investigate and determine the factors affecting vehicle pedestrian accidents taking place in busiest suburban highway Guilan Province located north Iran provide most accurate prediction model. Therefore, effective principal variables probability occurrence each category crashes are analyzed computed utilizing factor analysis, logit, Machine Learning approaches simultaneously. This method not only could contribute achieving comprehensive efficient model specify major contributing factor, but also it can officials with suggestions take measures higher precision lessen accident impacts improve road safety. Both analysis logit show significant roles exceeding lawful speed, rainy weather driver age (30–50) severity accidents. On other hand, lighting condition as severity, underline dominant role environmental all vehicle-pedestrian Moreover, considering both utilized methods, machine-learning has predictive power cases, especially accidents, 41.6% increase fatal 12.4% whole Thus, Artificial Neural Network chosen superior approach predicting number crashes. Besides, good performance validation machine learning proved through sensitivity analysis.