Cell Phone Distraction: Data Mining Application on Fatality Analysis Reporting System (FARS)

Distracted driving has become one of the ubiquitous concerns in terms of traffic safety as the presence of portable technology and its emergence while driving considerably increase. The objective of this thesis was to investigate the relationship between this prevalent distraction and motor vehicle accidents and how cell distraction influences driver performance across the United States from years 2011 to 2015 by one of the most reliable databases the Fatality Analysis Reporting System (FARS). Applying data mining techniques was used to discover and explore the role of distracted drivers in fatal crashes. Classification algorithms utilized which are C5.0, C4.5 and Ctree decision trees in determining the most related variables for the manner of collision and predicting the vehicle collision patterns occurred in fatal accidents. In addition, study the most contributed attributes related to the most harmful event that happens due to cell phone distraction, classifying and predicting the most harmful consequences in fatal crashes for those distracted drivers. The results show that the driver-related factors’ contribution continuously increased in the five-year period to determine the manner of collision and the most harmful event regardless of the gender. The dangerous use of cell phone while driving was demonstrated at intersections and the contribution of the intersection to increase the likelihood to get involved in angle crash was illustrated. Additionally, most of the crashes for the distracted driver are not with a motor vehicle in motion due to the inability of maintaining lanes or improper lane change during driving task because the driver’s eyes and mind during the use of cell phone are off the road for extended periods of time. The results demonstrate that cell phone distraction does not just have dangerous impacts on drivers' lives or vehicles but on pedestrians as well. This thesis sheds light on a new cause of overcorrection and subsequent rollover. For a better understanding of the issue of cell phone use while driving, in order to create more precise data with respect to drivers' awareness to the cell phone use risk, it is essential to set up the extent of drivers’ cell phone use more precisely. In order to truly assess the share of cell phone distraction crashes in the total number of crashes, cell phone use ought to be recorded accurately in accident reports. Keywords: distracted driving, data mining techniques, driver-related factors, decision trees, overcorrection.