Applying Parallel Processing to Improve the Computation Speed of K-Nearest Neighbor Algorithm

K-Nearest Neighbor (KNN) is a widely used algorithm to gain an accurate and efficient classification. One of the drawbacks of the algorithm is the time required to calculate the distance for each point. In this paper, the aim is to speed up the KNN algorithm with the implementation of multi threads and multiprocessors to reduce the time to execute the algorithm. There are two medical datasets used to apply the KNN algorithm for the comparison of the sequential and parallel performance. The datasets utilized are Heart Test and Breast Cancer Wisconsin (Diagnostic) Data Sets. The multiprocessing is used to parallelize the computation of the distance of KNN. The parallel KNN outperforms the sequential version with the speedup increase, leading to reduction in time for both the processes and threads.