Towards the new Similarity Measures in Application of Machine Learning Techniques on Agriculture Dataset

k-Nearest Neighbor is a simple and effective classification method. The primary idea of this method is to calculate the distance from a query point to all of classified data points and make choice of a class which occurs maximum time in k closest neighbors. The Euclidean distance and cosine similarity the common choice for similarity metric among all the similarity measures. Apart from Euclidean and Cosine there are various similarity measures available and being used to calculate similarity in n-dimension vector space model for classification. Similarity calculation is complex operation and computationally need high time if vector dimension increases. Hence this paper explores the usefulness of nine different similarity measures in kNN and presents their experimental results on agriculture dataset. We also compared the time required to finish the classification task and concluded that Idivergence is taking minimum time compared to these algorithms. General Terms k-Nearest Neighbor, Similarity measure