There are two kinds of driving modes of float car at low speed. The misjudgement of these modes will affect the accuracy and efficiency of the calculation of float car real-time traffic conditions seriously. A SVM(support vector machine) based float car driving mode classification model was studied and designed, and a novel membership matrix-based feature evaluation and selection method was proposed. The classifier whose features are selected through this method made a great classification accuracy of 92.6% in test samples. The float car driving mode analysis enhances the accuracy of exiting system evidently.
Wu D D.[J].Zhu T Y, Lü W F, et al. A heuristic map-matching algorithm by using vector-based recognition Proc of the International Multi-Conference on Computing in the Global Information Technology (ICCGI’07). Piscataway: IEEE.2007,:-
Vapnik V N. The nature of statistical learning theory[M]. New York: Springer-Verlag, 1995:138-141
Langley P. Selection of relevant features in machine learning Proc AAAI Fall Symposium on Relevance. New Orleans: AAAI Press, 1994:140-144
Lee H M, Chen C M, Chen J M, et al. An efficient fuzzy classifier with feature selection based on fuzzy entropy[J].IEEE Trans on Systems Man Cybernet,Part B.2001, 31(3):426-432
Uncu O, Turksen I B. Two step feature selection: approximate functional dependency approach using membership values Proceedings 2004 IEEE International Conference on Fuzzy System. Piscataway: IEEE, 2004:1643-1648
Connon R L, Dave J V, Bezdek J C. Efficient implementation of the fuzzy c-means clustering algorithms[J].IEEE Trans on Patten Analysis and Machine Intelligence.1986, 8(2):248-255
Han J W, Kamber M. Data mining concepts and techniques[M]. Los Altos: Morgan Kaufmann, 2001:132-136