MyMediaLite
3.02
|
k-fold cross-validation split for rating prediction More...
Public Member Functions | |
RatingCrossValidationSplit (IRatings ratings, uint num_folds) | |
Create a k-fold split of rating prediction data. | |
Properties | |
uint | NumberOfFolds [get, set] |
The number of folds in this split. | |
IList< IRatings > | Test [get, set] |
Test data for the different folds. | |
IList< IRatings > | Train [get, set] |
Training data for the different folds. |
k-fold cross-validation split for rating prediction
Please note that k-fold cross-validation is not the best/most realistic way of evaluating recommender system algorithms. In particular, chronological splits (RatingsChronologicalSplit) are more realistic.
The dataset must not be modified after the split - this would lead to undefined behavior.
RatingCrossValidationSplit | ( | IRatings | ratings, |
uint | num_folds | ||
) | [inline] |
Create a k-fold split of rating prediction data.
ratings | the dataset |
num_folds | the number of folds |
uint NumberOfFolds [get, set] |