So where does machine learning factor into this? Well, technically, it’s perfectly possible to perform a form of predictive analytics without computers at all. You can, and probably often do, make predictions with a simple pencil and paper, or even just in your head.
Take an example of an independent store owner who wants to work out his most efficient opening hours. By looking at receipts of the store’s takings over the last few weeks, the owner could work out the revenue for each hour of the day. With this information, he could match it against running costs and fairly accurately predict what times to close up in the future to save money, without even needing a computer.
This is a perfect example of predictive analytics, however, it’s also a very basic one. Problems start arising when you want to scale the predictions and make them as accurate as possible. For instance, financial institutions and governments may be dealing with millions or even billions of dollars and enormous data sets. In cases like these, the power of computers and machine learning becomes a lot more relevant, not to mention more economical.