Create a data model of what you've learned

- You may not realize it, but you've practiced a form of human machine learning your whole life. Even though you don't call it that, you've created data models to help you make better predictions. A data model is an abstraction that helps you understand a larger data set. When you go to a hotel, they'll probably be unknown fixtures to turn on the bath or shower. You've probably never seen these fixtures before, but you know how to turn a few knobs to get water from the faucets. When you need to take a subway in a new city, you generally know how to get around. You'll need a card or a smartphone to ride. There'll usually be a map that shows where the train will run. You know that they'll often show up on a periodic schedule. In fact, your model might be so good that you don't have to speak the language. All these models help you make better predictions. You can predict what will happen when you turn the red knob on that shower. You can predict what will happen when a train pulls into the station. These data models help you move around the world. If you didn't have them, you'd have no way to make accurate guesses when you tried something new. Everything new would be completely unknown. As it turns out, machines also need these models. They just need to summarize large data into a smaller data model. That way they can make better predictions. Even though they go through a very different process, they still use these models the same way. If you sell products, you might have a machine learning algorithm go through all your purchasing data. The machine learning system will then create a data model of what most people buy. This data model will be a summary of everything that the machine learning algorithms have learned from the purchasing history. The system will find patterns and then use those patterns to make predictions. It's the same one that you saw. Patterns in hotel fixtures or patterns and public transportation systems. The data model will help your machine learning system better understand the larger data set, maybe sell more fitness equipment right after the new year. A lot of people make New Year's resolutions to start exercising more. You probably sell more sunblock in the summer or more sweaters in the winter. As you can imagine, a very fine tuned data model has a lot of business value. You can make better inventory predictions that way. You won't have to warehouse sweaters in the middle of the summer. You can also offer sales and promotions just when your customer is looking for that product. That's why data models have become a very valuable piece of business property. Your machine learning teams will always be working on improving the model. They'll always drive to improve its accuracy. If you think about it, that makes a lot of sense. Netflix wouldn't be a streaming service if it hadn't been for the recommendation engine. People use Spotify because it's very good at finding other music that they like. The ability to predict your preferences or anticipate your buying behavior is often at the very heart of a successful business.