Splits a single input dataset into training, validation, and testing subsets. The proportion of the training set is controlled by the user; the remaining data is split equally into the validation and testing sets.
The node also allows for stratified sampling based on a string or boolean column of the input data. The column contents must have a domain defined on them.
A seed value can be supplied for randomized sampling to allow for consistent results. If no seed is given, a random seed will be used each time.
Requires the "KNIME Streaming Execution" extension.
Options
Output Ports
- Training set. The size of this table is defined by the slider control in the configuration dialog.
- Validation set. The size of this table is half of the remaining data after the training set has been split off.
- Testing set. The size of this table is half of the remaining data after the training set has been split off.
Nodes
Extensions
- FeatureKNIME Base nodes
- FeatureKNIME Javasnippet
- FeatureKNIME Math Expression (JEP)
- FeatureKNIME Quick Forms
Links
- No links available