Thoughtful Machine Learning with Python

Chapter 5. Decision Trees and Random Forests

Every day we make decisions. Every second we make decisions. Our perceptive brains receive roughly 12.5 gigabytes to 2.5 terabytes of information per second—an impressive amount—but they only focus on 60 bits of information per second.¹ Humans are exceptionally adept at taking in lots of data and quickly finding patterns in it.

But we’re not so great under pressure. In Chapter 1 we discussed flight and how checklists have solved many of its problems. We don’t use checklists because we’re stupid; on the contrary, it’s because under stress we forget small bits of information.

What if the effects of our decisions were even greater? Take, for instance, classifying mushrooms in the forest. If you are lucky enough to live in a climate that supports mushrooms such as morels, which are delicious, then you can see the allure of going to find your own, as they are quite expensive! But as we all know finding mushrooms in the forest is extremely dangerous if you misclassify them.

While death by mushroom is quite rare, the effects are well documented. Death caps, Amanita phalloides, cause liver failure. On the other hand, if you were to eat a Psilocybe semilanceata by accident, you would be in for a trip. Also known as liberty caps, these are the notorious magic mushrooms that people ingest to experience psychedelic effects!

While we as humans are good at processing information, we might not be the best at making decisions that are objective. When ...