I know how to predict ratings in test set with TrustSVD algorithm.
However, I also want to know the fitted values in the train set. I realize that the model seems to filter duplicated u-i pairs appeared both in the train set and test set, so I cannot use the strategy of using the train set as the test set.
I know how to predict ratings in test set with TrustSVD algorithm.
However, I also want to know the fitted values in the train set. I realize that the model seems to filter duplicated u-i pairs appeared both in the train set and test set, so I cannot use the strategy of using the train set as the test set.
So does anyone know how to achieve this goal?