For which type of model must labels be numeric?

In the context of machine learning and model types, regression models specifically require numeric labels because they are focused on predicting a continuous outcome. In regression tasks, the goal is to estimate a numerical value based on input features; for instance, predicting sales prices based on various factors like location and size of properties.

The requirement for numeric values arises from the nature of linear relationships that regression models analyze. The algorithms used for regression, such as linear regression or polynomial regression, rely on numerical calculations to identify the relationship between input variables (features) and output variables (labels).

On the other hand, classification models deal with categories or labels that can be either numeric or non-numeric. Clustering models focus on grouping data points based on similarity without the necessity of labeled outcomes. Object detection models utilize bounding boxes with coordinates and may involve classification labels, but again, these do not strictly require numeric outcomes across the board. Thus, the characteristic of requiring numeric labels distinctly identifies regression models in this context.