Analytical Skills for AI and Data Science

Appendix A. A Brief Introduction to Machine Learning

In this Appendix I briefly summarize what machine learning is with the purpose of providing a self-contained guide. I have not attempted to go into the intricacies of the methods, as the topic of the book is to learn to create value from these technologies, and not to learn each of these many different methods. It will provide some background knowledge and hopefully some intuition on how machine learning works. For interested readers, I will also cover the basics of A/B testing.

What Is Machine Learning?

Machine learning is the scientific discipline that studies ways that machines learn to accomplish certain tasks by using data and algorithms. Algorithms are recipes or sequences of instructions that are applied over and over until a precise objective is attained, and are written with programming languages that enable human interaction with computers. These are then translated to machine language that can then be processed and computed.

A Taxonomy of ML Models

Any taxonomy of machine learning algorithms starts by describing supervised and unsupervised methods (Figure A-1). In general, learning is supervised when someone or something tells us when the task was completed successfully. For instance, if you are learning to play a musical instrument, say the conga drums, your teacher might show you first how a good slap tone sounds. You try it yourself and she tells you whether the technique and sound were close to a perfect tone; ...