Metro traffic route assignment using K-Means clustering

Currently, many techniques have been applied to metro traffic route assignment, however all considering only priori probabilities. This paper presents a novel traffic assignment pattern, which unlike the conventional Logit-Dial algorithm. It introduces the Class Conditional Probabilities based on the empirical origin-destination (OD) data from Beijing Metro Networks. Firstly, a union set of effective paths is defined and constructed. Secondly, we employ K-Means clustering technology to calculate the probability density function of each path based on the presumption that the data is in accordance with Lognormal Distribution. Finally, given an OD record with specific travel time, we calculate its class conditional probabilities on each path, and assign the record to the path with maximum possibility. Experimental results show the correctness, accuracy and effectiveness of the proposed metro route assignment model.