DATA MINING AND MACHINE LEARNING APPROACHES IN DATA SCIENCE: PREDICTIVE MODELING OF TRAFFIC ACCIDENT CAUSES

Abstract

Due to the increasing number of deaths and injuries in traffic accidents today, it has become necessary to examine the potential contributing risk factors. The increase in the number of vehicles today leads to an increase in traffic accidents and loss of life and property. Analytical models are presented to investigate the socio-economic, demographic and temporal effects of the factors affecting the level of injury resulting from traffic accidents. By examining the data of various traffic accidents and developing a model, the factors and hazards affecting traffic accidents can be determined by data mining and machine learning approaches. The aim of this study is to determine which classification techniques are important for analyzing traffic accidents and to find out the factor that affects traffic accidents among the variables used in the research. The "Random Forest" algorithm, which gives the best model result among the techniques used in the research, was found. Weather conditions were found to be the most important factor among the factors that lead to traffic accidents, followed by the age and education of the driver. This study is a traceable application in terms of revealing the differences between data mining and machine learning and following the processes.