Parameterization of traffic flow using Sammon-Fuzzy clustering

Modelling the traffic conditions has become necessary in the modern connected society. We have attempted to use clustering algorithms to classify traffic flow in and around Pune city into classes representing geographical locations of sampling of the data. The algorithm employs Sammon´s mapping along with fuzzy clustering algorithms to cluster the data. Such high-end parameterization of traffic flow can help in better control and real-time modelling methods. The algorithm is applied to two different databases - traffic inside the city and traffic outside it and approximately 95% accuracy is obtained across vivid conditions.