Data Smart, 2nd Edition

6Regression: The Granddaddy of Supervised Artificial Intelligence

Today, we're bombarded with messages about the power of artificial intelligence (AI). AI can really do anything, we're told. Big companies tell us of an AI-powered future that doesn't seem too far off. Dazzling us with its potential, AI presents itself as something unobtainable without the help of a big company or tech or a chatbot.

But what if I told you that you could do start doing AI right now in Microsoft Excel?

Beyond those headline-catching, over-promised stories where AI can solve any problem, there is no doubt AI is useful. In fact, your credit card company uses it to identify odd transactions on your account, and the enemy in your shoot 'em up game runs on AI, as does your e-mail spam filtering, the bank's tax fraud detection, spelling autocorrection, and automated chat bot.

The public imagination and fascination with AI would have us believe that AI is an arcane set of rules understood by only the genius few; but in fact, as you'll see in this chapter, AI (and one of its primordial methods, linear regression) has been around for a while. Indeed, the original AI users didn't have boatloads of technology. Instead, they used what they had: historical data and the desire to learn from it. And their brains.

This book was originally published in 2013. The first edition mentioned many of the AI successes that stand as a precursor for how we live in AI today: the retailor Target made headlines for seemingly ...