Data Smart, 2nd Edition

8Forecasting: Breathe Easy: You Can't Win

As you saw in previous chapters, supervised machine learning is about predicting a value or classifying an observation using a model trained on past data. Forecasting is similar—past data is used to predict a future outcome. Indeed, some of the same techniques, such as multiple regression (introduced in Chapter 6, “Regression: The Granddaddy of Supervised Artificial Intelligence,”) are used in both disciplines.

But where forecasting and supervised machine learning differ greatly is in their canonical problem spaces. Typical forecasting problems are about taking some data point over time (sales, demand, supply, GDP, carbon emissions, or population, for example) and projecting that data into the future. And in the presence of trends, cycles, and the occasional act of God, the future data can be wildly outside the bounds of the observed past.

You see, that's the problem with forecasting: in previous chapters we saw the buying habits of pregnant women, who more or less buy the same stuff. But what if the future looked nothing like the past? How do your predictions account for the unpredictable? Future time-series data can and will look different than the data you've observed before.

Just when you think you have a good projection for housing demand, the housing bubble bursts. Your forecast is in the toilet.

Just when you think you have a good demand forecast, a global pandemic disrupts your supply chain, limiting your supply and forcing ...