Short-term traffic flow prediction based on ratio-median lengths of intervals two-factors high-order fuzzy time series

Abstract

Due to the complexity of traffic flow characteristics, the traditional statistical regression models have been unsuitable for the traffic flow prediction. And thereby the paper proposes the fuzzy time series method to predict short- term traffic flow. First, we proposes an improved fuzzy time series prediction model, i.e. , ratio-median lengths of intervals two-factors high-order fuzzy time series. The prediction model simultaneously considers impact of many factors on the traffic flow formulation. For achieving higher prediction accuracy, the ratio-median lengths of intervals method is adopted to adaptively partition the universe of discourse of linguistic variable. Then it is used to predict the raw traffic flow data which are collected at Zizhu Bridge in Beijing. The experiment result verifies that the improved fuzzy time series prediction model can achieve high prediction accuracy.