In this chapter, we created a basic but useful e-mail subject line tester. This chapter provided a guide to code a basic Naïve Bayes classifier from scratch without any external library, in order to demonstrate how easy it is to program a machine-learning algorithm. We also defined the maximum size threshold for the training set and got an accuracy of 92 percent, which, for this basic example, is quite good.
In the following chapters, we will introduce more complex machine learning algorithms, using the
mlpy library, and we will also present how to extract more sophisticated features.