Currently it is very difficult to work with tables that are generated from the mod port after an apply model operator. It would be convenient for example to be able to generate an exampleset from the weights of a linear SVM or the Support Vector table. Right now the only way to use the weights from LibSVM is to 'write as text' and then parse the file. A simple idea like trying to sort the weight table becomes quite complex. If the output is from a group models operator then managing the data is even more problematic.
Since we have weights to data, performance to data and similarity to data operators already in place, something to work with models would be very handy.
Alternatively, having a weights port on the apply model operator would also do the job as we could then use an existing operator to output the exampleset.