Specific implementation of Cross-Validation modeling when:
- a Source Dataset is being numerously split into Training dataset and Validation dataset
- each realization of Source Dataset split is used to train the model and estimate its discrepancy
- the model calibration continues until the average discrepancy on all Source Dataset split realizations stop decreasing
This ensures that a final model has the best predictability against random pick on validation data points.
The number of Source Dataset split realizations and percentage of Training dataset / Validation dataset data points are set manually.
See also
Natural Science / System / Model
[ Cross-Validation ]
Formal science / Mathematics / Statistics