Application of classification, clustering and prediction algorithms in the detection of patterns associated with mobility using vehicle trajectory data

Abstract

Context: Traffic problems represent an impediment to the personal development of students and workers who must meet specific schedules. This article seeks to deepen the study of this problem and serve as a precedent for future research. Method: Classification (K Nearest Neighbor), clustering (K-mean) and prediction (Linear Regression) algorithms were applied to a database of vehicle trajectories, using three datasets with information on distance, duration, temperature and time of day. Results: A relationship is found between high temperatures and longer trip lengths, trips with the same distance but different durations, and longer durations at midday and in the afternoon. Conclusions: The relationship between high temperatures and longer trip lengths may be due to mobility problems due to high traffic volume in the midday hours. The differences in travel time for trips of equal distance could be explained by the routes and times at which users made them. Finally, this study lays the groundwork for future research that seeks to analyze and establish solutions to the traffic problems that affect the personal development of students and workers.