In this chapter, we started with a discussion of supervised and unsupervised learning and emphasized the difference between pure predictive and exploratory analytics. We were then introduced to the first of the core algorithms (general linear models) which are important in the predictive analytics world. We then discussed various regression methods, along with its pros and cons, and noted that regression can be an extremely flexible and well researched statistical based modeling tool. We then used a pain threshold study to show examples of logistic regression and regularized regression, along with discussing important regressions concepts such as interaction, p-values and effect sizes.
In the next chapter, we will resume our discussion of the core predictive analytics algorithms by discussing three additional algorithms, that is, decision trees, clustering, and support vector machines.