The cross-validation devides the data into n folds and measures the accuracy n times therefore. The displayed RMSE in the Test & Score tool is a mean of all n runs, I think. But is it possible to get the mean coefficents of regression as well?
At the moment my process looks like this:
1. Loading data from a csv with one target and some features. All are numerical data.
2. Rank features. Use the most important two features
3. Use Test&Score with a linear Ridge Regression learner to quantify the regression. I'm using the cross-validation with ten folds in the Test&Score tool.
In this case the data table shows the coefficients independent from the Test & Score tool. Even if I delete the Test & Score thing, there will be no changes on the regression coefficients.
thanks for Your help