Aggregates the predictions per row to be explained and calculates the Shapley Values for each feature prediction combination. For each explained row (top inputs of the Shapley Values Loop Start node), this node outputs number of prediction columns rows where each row contains the explanation for one prediction column. This kind of explanation row consists of the four columns "RowId", "Target", "Actual prediction", "Deviation from mean prediction", and a row for each feature (selected in the Shapley Values Loop Start node). The meanings of these are as follows:
- RowId: The RowId of the row of interest.
- Target: The name of the prediction column that is explained.
- Actual prediction: The actual prediction your model created for the row of interest.
- Deviation from mean prediction: The difference between the prediction of the row of interest and the mean prediction of the model. The mean prediction is estimated from the model's predictions on the sampling table (bottom input of the Shapley Values Loop Start node).