Classification of Cellular Phone Mobility using Naive Bayes Model

Abstract
Road traffic data is a fundamental element of the intelligent traffic system (ITS). However, the availability of road traffic data is currently limited due to the high investment of traffic sensors and associated infrastructure. Using cellular phone information as road traffic data becomes an attractive alternative because of its low cost, widespread of cellular networks, and a large number of phones as potential road traffic probes. However, in practice, the collected cellular data consists of various types of mobility, either related or unrelated to the road traffic. In this paper, we proposed a method to classify two types of mobility, i.e., sky train and pedestrian, from cellular phone information. Two key attributes, i.e., 1) the number of unique cell ID and 2) the average cell dwell time of unique cell ID are used in Navies Bayes classification model. The experimental results show promising performance with accuracy up to 93.1%. This suggests a potential use of cellular phone information as road traffic data.

This publication has 8 references indexed in Scilit: