Modern regression methods with applications

Abstract

Nowadays more and more scientists, corporations, employees and academic students around the world use statistical analysis and various regression models in order to analyze different datasets and come to conclusions that will help them advance in their jobs or even in their lives. Just like all these people, in this specific thesis report, we are also going to use statistical tools like R programming language in order examine eight different regression methods (multiple linear, ridge, lasso, polynomial, logistic, poisson, decision tree and random forest) over a dataset that contains information about a variety of Portuguese red wines. Through this pioneer project we have faith that we’ll find out the most suitable regression type/s for a wine quality analysis, but also for a discrete dataset analysis generally. In the following sections we will present and analyze the theoretical background for each one of the eight different prediction methods, a significant amount of literature review that exists at the moment and which is dedicated to the practical application of these models over a variety of real-life sectors and of course we are going to compare the statistical results that we are about to receive from the usage of Rstudio and that will help us succeed in our goal, shedding light on which is the most appropriate regression type for this specific task.