just i would like to know is there any way to appending two same length column without changng the row ID? it means that the output should have an original row ID of the first column before appending for instance. Because i have a cross validation loop in my workflow so i use a joiner to join column by row ID in middle of workflow but in each loop of cross validation since the row ID of the joiner output is the same i've got an error :
ERROR X-Aggregator Execute failed: Encountered duplicate row ID "1" at row number 22
Have you tried the column appender node.
yes i have but coulmn appender is also similar to joiner so the row ID should be same to join them.
the problem is just the cross validation loop all the things are ok. Its actually odd why the cross validation aggregator work in such a way which in each loop the row ID of prediction and target matrix should be different.
I believe when using the x validation looping, only the rows from the 2nd output of the cross partitioner node should be fed into the cross validation aggregator node.
basically the top outport rows should be used to build a model, whilst the second port rows are used to get predictions on, these are then fed to the aggregator for analysing the errors.
does this make sense.
Thanks Simon i've used another approach and it works finally. I used CSV writer to write the test matrix in to the csv with original Row ID and i retrieve it and combine it by cbind command with the prediction matrix of my model in R snippest. By using this way the matrix row ID which fed in to the aggregator didnt change anymore in each loop.Take a look on workflow.