Comparative study of K-NN algorithm for transportation mode detection using mobile phone sensor data

This paper examines the use of mobile phone sensor data to identify transportation mode detection using the K-nearest Neighbor algorithm. The model tries to recognize the walking, still walking, Bus, Train, and Car transportation modes. One of the normalization methods, such as the Min-Max normalization or the Z-Score Normalization, is implemented to pre-process the data. It uses four distance methods such as Euclidian, Manhattan, Chebyshev, and Minkowski as distance calculation mechanisms. Based on the highest accuracy result, the model is selected. The analysis also concluded that the optimal model has the highest accuracy, which has validated the results achieved through extensive normalization methods and the choosing of the most appropriate distance functions. As a result, the outcomes of the research have further indicated the importance of selecting the appropriate normalization techniques and distance functions on the accuracy of the model used in transportation detection. Additionally, the results have also provided critical additional knowledge in the development and formation of Intelligent Transportation Systems (ITS).