ABSTRACT
Classification algorithms learn by examining already existing data, classified into various classes and create models which are used to find a predictive pattern. This paper aims at studying the efficiency of an existing classification algorithm, k-Nearest Neighbours (kNN) and suggest a new, space optimized approach of the algorithm. The purpose of this research is to reduce the space required by kNN, thus reducing the execution time for larger datasets and improving the accuracy simultaneously. The aim is to do so by removing irrelevant features and reducing the amount of data required to train the model.
