The first rule of machine learning is seeing its use in everyday life. From deciding which car is better based on features rather than pictures of it, to deciding based on pictures whether a robot is folding clothes correctly, to learning to hear by modeling the auditory function of the brain to turn sound waves into text representations your brain can understand, machine learning is everywhere!

To do machine learning, you need data—many times, lots of it, but not all the time. That data is often not prepared the right way and needs to be cleaned up. You’ll also need some suppositions or hypotheses about that data that you’d like to test and evaluate. Speaking of which, you also need tools to evaluate how well your machine learning is making predictions, or grouping things, or ranking and rating things. All these components are part of your machine-learning rig, which you will use to take in input data and answer questions about that data in meaningful ways that you can evaluate.

The first part of the book focuses on the components of the machine-learning rig, showing how Google’s TensorFlow framework, along with a set of associated machine-learning tools and utilities in the Python programming language, can help you apply and use the machine-learning rig for the real-world problems that you’ll explore in the remainder of the book.