A class label is a categorical value assigned to an observation to identify its class. When that class is what a model aims to predict, the label is the outcome, or dependent variable. Predictors are the inputs used to estimate it. section 9.1 section 9.1.2
Not every category in a dataset is an outcome label. Treatment and control identify comparison groups; an experiment measures an outcome to compare those groups. Likewise, a high-versus-low precipitation indicator can be an input to a model. Its categorical values do not make it the outcome. section 3.2.4 section 9.3.1
Decide which variable answers the question before assigning modeling roles. A model built to predict a category calls for a class label; the worked housing-price model instead estimates a numeric outcome. Classification is noted, but a class-label prediction workflow is not worked through here. section 9.1.2 section 9.1
