But where forecasting and supervised machine learning differ greatly is in their canonical problem spaces.

business receives feedback on how their audience is engaging at the individual level through click tracking, online purchases, social sharing, and so on.

You'll be going over a type of ensemble model called bagged decision stumps, which is very close to an approach used constantly in industry called the random forest model.

just cleverly turns past data into a formula or set of rules that it uses to predict a future case. As we saw in the case of naïve Bayes in Chapter 3, it's the AI model's ability to recall this data and associated decision rules, probabilities, or coefficients that make it so effective.

This story about Target is just one of many that have peppered the press recently. Watson won Jeopardy!. Netflix offered a million dollar prize to improve its recommendation system. The Obama re-election campaign used artificial intelligence to help direct ground, online, and on the air media and fundraising operations. And then there's Kaggle.com,

Most predictive model performance metrics are based on counts and ratios of four values that come from the predictions on our test set:

Using k-means clustering, you can find the best segments and then try to understand why they're the best segments.

You get more bang for your buck spending your time on selecting good data and features than models.

At a simplistic level, you feed a supervised AI algorithm some historical data, purchases at Target for example, and you tell the algorithm, “Hey, these purchases were from pregnant people, and these other purchases were from not-so-pregnant people.”