A comparison of classification algorithms

Abstract

Classification is a important problem in statistical and machine learning. In this thesis, I compare four widely used classification methods called Naïve Bayes, K-Nearest Neighbors, Logistic Regression, Linear and Quadratic Discriminant Analysis. These methods are compared in terms of classification accuracy and prevision. The thesis is organised in three chapters. The first chapter describes the methodology. The second chapter illustrates the methods with simulations. The final chapter demonstrates the performance of the classification methods with an application to a real dataset. Computations are performed in the R language.