In this chapter, we iteratively developed the foundation of a neural network (a perceptron) using TDD by first developing concrete scenarios in a spreadsheet. Using a spreadsheet allowed us to get a handle on each of the different calculations that are required in a perceptron. We also saw that we can use TDD without needing to prove the mathematics. Instead we used TDD to implement the smallest amount of the mathematics that we needed to make our perceptron perform incrementally better. A side benefit to this is that you may have also developed a deeper understanding of what each aspect of the perceptron actually does.

In the next chapter, we will continue exploring mathematical solutions to machine learning. Specifically, we will develop a solution to maximize the money we can make when we have several different options of what to show customers and have no idea which is the most profitable.