package jsat.datatransform;
import java.io.Serializable;
import jsat.DataSet;
import jsat.classifiers.DataPoint;
import jsat.exceptions.FailedToFitException;
/**
* A pre-processing step may be desirable before training. If a pre-processing
* step is used, it is necessary to also apply the same transform on the input
* being sent to the learning algorithm. This interface provides the needed
* mechanism.
* A transform may or may not require training, it could be fully specified at
* construction, or learned from the data set. Learning is done via the
* {@link #fit(jsat.DataSet) fit method}. Many DataTransforms will include a
* constructor that takes a dataset as a parameter. These transforms will fit
* the data when constructed, and exist for convenience.
*
* @author Edward Raff
*/
public interface DataTransform extends Cloneable, Serializable
{
/**
* Returns a new data point that is a transformation of the original data
* point. This new data point is a different object, but may contain the
* same references as the original data point. It is not guaranteed that you
* can mutate the transformed point without having a side effect on the
* original point.
*
* @param dp the data point to apply a transformation to
* @return a transformed data point
*/
public DataPoint transform(DataPoint dp);
/**
* Fits this transform to the given dataset. Some transforms can only be
* learned from classification or regression datasets. If an incompatible
* dataset type is given, a {@link FailedToFitException} exception may be
* thrown.
*
* @param data the dataset to fir this transform to
* @throws FailedToFitException if the dataset type is not compatible with
* the transform
*/
public void fit(DataSet data);
public DataTransform clone();
}
Ads help maintain this website.