I’m a beginner on KNIME and I want to know how can I replace the missing values in a table with the values that predicted through a predictive model? I’ve done following:
- filter out the rows with missing values for testing
- train the complete data - loop over the columns with missing values (each loop one column for predicting)
- test on the row with missing values and get the predict value
now I need to replace the missing value in the rows with the predicted values and make the original dataset complete…
or are there any better idea for the predicting process? Thanks.
I’m not sure about the structure of the data you end up with after predicting the missings. But one thing you could do is to use a Column Merger node. This node merges two columns into one by choosing whichever cell is non-missing.
Hope that helps!
How do you want to evaluate performance of your predictive model ? Test set should contain real output values (not missing) which have to be compared with those coming from predictive model.
I just want to use a predictive model for predicting the missing values in the data set and replace the missing values with the predicted value. (in a cross validation, as pre-processing process)