Modeling and Multivariate Methods - SAS

Chapter 13 Recursively Partitioning Data 329
Partition Method
Bootstrap Forest
The Bootstrap Forest method makes many trees, and averages the predicted values to get the final predicted
value. Each tree is grown on a different random sample (with replacement) of observations, and each split
on each tree considers only a random sample of candidate columns for splitting. The process can use
validation to assess how many trees to grow, not to exceed the specified number of trees.
Another word for bootstrap-averaging is bagging. Those observations included in the growing of a tree are
called the in-bag sample, abbreviated IB. Those not included are called the out-of-bag sample, abbreviated
OOB.
Bootstrap Forest Fitting Options
If the Bootstrap Forest method is selected on the platform launch window, the Bootstrap Forest options
window appears after clicking OK. Figure 13.10 shows the window using the Car Poll.jmp data table. The
column sex is used as the response, and the other columns are used as the predictors.
Figure 13.10 Bootstrap Forest Fitting Options
The options on the Bootstrap Forest options window are described here:
Number of rows
Number of terms
gives the number of observations in the data table.
gives the number of columns specified as predictors.
Number of trees in the forest
is the number of trees to grow, and then average together.
Number of terms sampled per split is the number of columns to consider as splitting candidates at
each split. For each split, a new random sample of columns is taken as the candidate set.
Bootstrap sample rate is the proportion of observations to sample (with replacement) for growing each
tree. A new random sample is generated for each tree.
Minimum Splits Per Tree
is the minimum number of splits for each tree.
Minimum Size Split
is the minimum number of observations needed on a candidate split.