Supervised machine learning models based on support vector regression

Abstract A vast majority of practically utilized machine methods in fact involve supervised learning techniques. In supervised learning, an algorithm is used to learn an approximate mapping function from the input variable x to the output y = f ( x ) so that when there is new input data, we can predict the output for that data. Support vector regression, which is derived from its parent version, is based on an optimization problem. The weights corresponding to each input sample in a training set are obtained. Variants such as least square support vector regression, twin support vector regression, and e-twin support vector regression are presented with a case study for each. In this chapter, such regression models are described in detail.