Parameters | Predictive Modeling | Method
Method
Use the drop-down menu to specify the method by which to hold out test observations for cross validation.
Available methods for holding out test observations are described in the table below:
Method
Choose this method to...
Simple Random
...hold out a fraction of the data selected completely randomly.
Stratified Random
...randomly hold out a fraction of the data selected randomly within each trait group.
Random Partition
...randomly partition all of the data into groups and hold out each group in turn.
Block Partition
...partition all of the data into blocks of consecutive observations and hold out each group in turn.
Split Partition
...partition all of the data into k parts, with the ith part consisting of observations i, i + k, i + 2k, ....
All Possible Subsets
...hold out all possible subsets of a certain size.
Currently only subsets of size 1, 2, or 3 are available.
To Specify a Method for Holding Out Observations:
*
Select the desired method using the drop-down menu.