Passing undefined classes to classification algorithm using row splitting

Hi everyone, please help!

I have a set of around 2,500 (pre-processed) free text comments, the first 10% have been classified. I have used the sorter and row splitter nodes to isolate the classified comments from the unclassified ('undefined') ones - and a decision tree model has been trained on the 250 or so comments I know the class for with a decent level of accuracy ..... so far so good.

My problem is that I now have this lump of over 2,000 undefined comments which need classifying using the aforementioned DT model. How do I get it to replace the class labels of the comments we don't know the sentiment for from 'undefined' to one of the 3 classes (pos, neg, mix) identified in the first 10% of comments?

I'm sorry if this has been done to death, but I'm having real difficulties getting this to work. I have attached a screen shot of my workflow below.

Thanks so much in advance for your help. It's much appreciated :)


Just remove the column with 'undefined' values and set the Decision Tree Predictor to output the column with the original column name.

Cheers, gabor

Thanks Gabor. I'll give that a go.