Performance Analysis of Machine Learning Algorithms on Small Datasets that Includes Features from K-Nearest Neighbor Graph

Abstract

Modern technology in today’s world is largely driven by machine learning algorithms. They are incorporated into every field. Big data is not always available to us, though. We frequently have to work with limited-size of data. The purpose of this paper is to demonstrate several machine learning algorithms and their accuracy on small numerical datasets. We investigate the effectiveness of these algorithms with and without the implementation of two variables, degree and closeness centrality, which are extracted from the dataset using the knearest neighbor graph.