difference in R squared values from Regression vs Scorer node

What is the difference between the R-squared statistic given by the Linear Regression learner, and the one from the numeric scorer node? They don't seem to agree...

here's what I've done: I have a dataset with 3 variables and one target, all numeric. I run it through a Linear Regression Learner node, with a predefined offset of 0.

The Result View shows a R squared of 0.7094

I use the generated model with the Regression predictor node on the same data, and run the output through the numeric scorer node.

The statistics view shows a R squared of 0.6201

How is this possible? (and more importantly, which one is correct!!??)



I think I have found the answer to this myself. (It's friday, which is a good excuse for not having thought this through before posting :/ )

The scorer node calculates the coefficient of correlation between the actual and predicted values.

The regression learner calculates multiple R-squared, which is defined as the square of the coefficient of multiple correlation, it takes into account the correlation of each of the variables to the target. The scorer node does not.

(it would still be nice if someone could confirm this :) )