Regression Tasks For Machine Learning

Abstract

Key aspects of machine learning include predictions and classifications, then detection and tracking of the objects and the environment to finally capture the data and adapt as needed. This chapter will introduce the theoretical aspects of regression from simple to multilinear models and addresses to tackle the bias and variance to a certain extent. Regression task in machine learning is a method for prediction of a continuous variable which is a dependent variable. Regression techniques fall under the category of supervised learning. Generally, regression models are based on the relationship between the dependent variable and the set of independent variables. Regression models are applied in various domains such as healthcare predictions, forecasting stock prices, house prices, and in trend analysis. In the machine learning context, regression models are used to fit the data points along a line as a best fit and minimize the distance between the data points and the line by least squares methods. This chapter begins with a simple linear regression and diagnosis and then how to select features from the given set of independent or predictor variables, importantly the utilization of squared R (coefficient of determination), p-values and F-score. To understand the relationship between the dependent variable and the set of independent variables, various visualization methods are discussed in this chapter. Also, this chapter deals with statistical modeling via data visualization, with the help of visualization, and the diagnosis of the model. © 2023 Elsevier Inc. All rights reserved.