data member of the object contains always the following columns:
id, index numbers of predicted cases from the task,
either a numeric or a factor, the predicted response values,
either a numeric or a factor, the true target values.
If probabilities were predicted, as many numeric columns as there were classes named
prob.classname. If standard errors were predicted, a numeric column named
makePrediction is mainly for internal use.
Type set in setPredictType.
Threshold set in predict function.
Task description object.
Time learner needed to generate predictions.
Any error messages generated by the learner (default NA_character_).
Internal, do not use!
makePrediction( task.desc, row.names, id, truth, predict.type, predict.threshold = NULL, y, time, error = NA_character_, dump = NULL )