Introduction

The aim of a supervised machine learning method is to build a learning model from training data that includes input data containing known labels or results. The model can predict the characteristics or patterns of unseen instances. In general, the input of the training model is made by a pair of input vectors and expected values. If the output variable of an objective function is continuous (but can also be binary), the learning method is regarded as regression analysis. Alternatively, if the desired output is categorical, the learning process is considered as a classification method.

Regression analysis is often employed to model and analyze the relationship between a dependent (response) variable and one or more independent (predictor) variables. One can use regression to build a prediction model that first finds the best-fitted model with minimized squared error of input data. The fitted model can then be further applied to data for continuous value prediction. For example...