A new outlier detection algorithm and its application in intelligent transportation system

Abstract

Outlier detection plays an important role for data analysis in data mining. Aiming at outlier characters of Intelligent Transportation System (ITS) such as few samples, high frequency and large range, a new outlier detection algorithm based on probability theory and fuzzy clustering method (FCM) is proposed. Firstly, the new algorithm judges data variation, and then clusters data using FCM. Finally, the outlier detection result is given through estimating clustering result using probability theory. Detection of practical travel time verifies validity and practicability of the new algorithm.