Congestion has become a major threat to a country's economy. It not only causes loss in terms of man-hours and fuel costs, but also causes frustration among the public. It has become important for traffic operators to clear off congestion in a timely manner to resume the normal flow of traffic. This research focuses on evaluating congestion detection algorithms to implement them in an Intelligent Transportation System to detect congestion in real time. We have evaluated statistical based algorithms as well pattern recognition based algorithms to detect non-recurrent congestion on I-74, Davenport, Iowa, USA. Inter-Quartile Distance based algorithms and Supervised Learning based Decision Tree and Random Forest Classifiers are compared and evaluated in this study.