There are known knowns; there are things that we know that we know. We also know there are known unknowns; that is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don’t know we don’t know.
Classification and regression are powerful, well-studied techniques in machine learning. Chapter 4 demonstrated a classifier as a predictor of unknown values. There was a catch: in order to predict unknown values for new data, we had to know that target value for many previously seen examples. Classifiers can only help if we, the data scientists, know what we are looking for already, and can provide plenty of examples where input produced a known output. These were collectively known as supervised learning techniques, because their learning process receives the correct output value for each example in the input.
However, there are problems in which the correct output is unknown for some or all examples. Consider the problem of dividing up an ecommerce site’s customers by their shopping habits and tastes. The input features are their purchases, clicks, demographic information, and more. The output should be groupings of customers. Perhaps one group will represent fashion-conscious buyers, another will turn out to correspond to price-sensitive bargain hunters, and so on.
If you were asked to determine this target label for each new customer, ...