Performance Measures for Short Term Traffic Condition Forecasting Algoithms

Short term traffic condition forecasting has been identified as one of the critical issues to be tackled for developing proactive traffic management and control systems. Traditionally, the accuracy measures are used to gauge the performance of the forecasting algorithm. In this paper, five measures of accuracy, transferability, efficiency, adaptability, and robustness are proposed and discussed for supporting the construction of the real world applications.