Uma análise do algoritmo K-means como introdução ao Aprendizado de Máquinas.

Abstract

This work aims to analyze the convergence of the K-means method, an unsupervised learning algorithm that groups n data into k-clusters. In this sense, we presented some of the advantages and disadvantages of the K-means method, comparing the original clustering and the clustering done by the algorithm. Also, we presented the application of the algorithm in two data sets: breast cancer and diabetes, analyzing the clustering done by K-means as well as the patterns and regularities present in the clusters. In this way, we seek to present an introductory study of Machine Learning theory, which seeks to make machines perform tasks without being instructed all the time, starting only from some initial instructions. Specifically, we seek to understand some of its definitions and characteristics that will allow identifying the type of learning studied.