The mean is the arithmetic average: add the values and divide by how many values there are. In regression, it gives a simple prediction for a group by placing a horizontal line at the labels’ average. section 9.14
A small regression example
The eight engagement labels are 7, 5, 7, 1, 2, 1, 5, and 4. Their mean is 4, so a model that gives every user the same prediction predicts 4. After splitting users younger than 35 from the rest, the group means are 6.333 and 2.6.
Regression trees use means in a repeated process:
- Split the data using a feature cutoff. section 9.14
- Predict each resulting group with its labels’ mean. section 9.14


































