Data Analysis Applied to Driver Profile Data Obtained Through Vehicle Telemetry and Classification Algorithms

Abstract

In the present days, considering the enormous amount of unused data available, one of the most challenging tasks is extracting meaningful information from data obtained or stored; for the automotive industry, where communications systems can transmit a large amount of data, such data analysis could be used to improve the vehicle’s or the driver’s performance. Beyond receiving and treating data, the results and the outcome of how the data is used for in-depth analysis bring out the work’s value. Through data analysis and visualization, several views were made with the data obtained after using an algorithm of classification for driver behavior profiles previously created by the authors. The data of 5151 trips were acquired; 3088 were deemed suitable to use and within the parameters (59.95%). Split between economic and behavior profiles, seven profiles of drivers were used, 3 for economic profiles (out of which 13.5% were economic, 82.55% were average, and 3.95% were big spenders), 3 for behavior profiles (28.76% cautious drivers, 55.44% average drivers and 14.54% aggressive drivers), and an extra profile (1.26%) was used for exceptional cases when the vehicle stopped for a long time.