Systematic identification of peak traffic period

To identify peak periods in systematic manner, this paper proposes and tests a framework to pinpoint the periods by mining travel time time-series data. In differentiating between peak and off-peak periods in the data, Bottom-Up algorithm, a method of segmenting time series, is presented. Each day’s travel time time-series is segmented and the time points of the start and the end of the peak period is identified. These points (start and end of the peak) are obtained for different days of interest and the respective distribution is defined to estimate a statistically significant time for start and end of the peak for the site. The applicability of the proposed framework is tested using real Bluetooth data from Brisbane. The case study analysis indicates that the peak periods can be systematically estimated using the proposed framework and the results are better than the existing threshold based approach. The paper also presents practical applications that can be improved by implementing the proposed method.