A Recommendation System based on Clustering and Classification for Optimal Trajectory Analysis

Abstract

Moving objects such as people, animals, and vehicles have generated a huge amount of spatiotemporal data by using location-capture technologies and mobile devices. There is a high demand to analyze this collected data and extract the desired knowledge. In this study, we built a recommendation system based on four data mining techniques which are clustering, classification, sequential pattern mining, and time series analysis. We have focused on predicting traffic status in an effective way by considering the trip destination which can be useful for passengers. We applied clustering and sequential pattern mining to detect taxi trips movement in different areas, then we applied the Naïve Bayes classifier to predict the traffic status of each trip. With the real taxi trips data of 441 taxis, we performed qualitative and quantitative analysis for our clustering method, then we evaluated the accuracy of the classification models. The results show that our recommendation system can achieve 70% accuracy in predicting traffic status.