Understanding Regression
Machine learning deals with supervised and unsupervised problems. In unsupervised learning problems, there is no historical data that tells you the correct grouping for data. Therefore, these problems are dealt with by looking at hidden structures in the data and grouping that data based on those hidden structures. This is in contrast with supervised learning problems, wherein historical data that has the correct grouping is available.
Regression is a type of supervised learning. The objective of a regression model is to predict a continuous outcome based on data. This is as opposed to predicting which group a data point belongs to (called classification, which will be covered in Chapter 7, Predicting Customer Churn). Because regression is a supervised learning technique, the model built thus requires past data where the outcome is known, so that it can learn the patterns in the historical data and make predictions about the new data. The following figure illustrates...
