Help中的内容如下:
Choose the data Role from the drop-list. The role determines how the data set is used throughout the process flow. The role may be any of the following:
- RAW - used as raw input to the node (default).
- TRAIN - used to fit initial models.
- VALIDATE - used by default for model assessment. The VALIDATE data set is also used for fine tuning the model. Note: The Decision Tree and Neural Network nodes have the capacity of over fitting the TRAIN data set. To prevent these nodes from over fitting the TRAIN data set, the VALIDATE data set is automatically used to retreat to a simpler fit than the fit based on the TRAIN data alone. The VALIDATE data set can also be used by the Regression node for fine tuning stepwise regression models.
- TEST - additional "hold out" data set that you can use for model assessment.
- SCORE - used to score a new data set that may not contain the target.
******************************************8
但是这些role的用法究竟如何?如何影响后续的节点?