Creating weka_results, please wait ...
Hi everyone
Whenever I create a model using Weka operators available through Weka extension, I'm not able to see the model itself. So I connect the model port to the results port and after the process is done, there are to tabs, one is called 'description' and the other is 'weka results' which is supposed to show the actual model. But it says "creating weka_results, please wait ..." and never shows me the model.
See the attachment for more information.
I'm using RapidMiner 7.0 on OSX El Capitan if that matters.
Answers
-
hmm that's strange. I use Weka models all the time and they show up. I would first try upgrading to RM 7.3 and ensuring that your Weka extension is also up-to-date.
Scott
0 -
I've noticed that sometimes, depending on how much data the Weka model crunches, showing the model might take a long time. I would definately use v7.2 and enable the Beta mode. We're redoing the data core of RapidMiner to make things more effecient and speedier.
1 -
I upgraded my Rapidminer Studio as well as the Weka extension to 7.3, still the same problem exists.
Also I'm using Rapidminer 7.0 on a Ubuntu machine, and just installed 7.3 on a windows 10 machine, didn't work on those two either ...
0 -
Also enabled the Beta mode, didn't help ...
To make sure that the problem is not as a result of a large data, I created a simple process with few data instances and the message still shows up and never goes away.
0 -
What kind of Weka alogrithm is it?
0 -
Initially it was RandomForest, then tried logistic regression and even OneR, still the same.
I would like to remind that the 'description' tab of the model results work, and this is what happens with the 'weka results' tab.
0 -
0
-
Did you try running your setup through the Weka program. Maybe it's just how the extension is implements their libriaries. I couldn't say, I rarely use the Weka extension.
0 -
I ran the setup through Weka program and it outputs the model (though not in a nice graphical way like what RapidMiner does). The thing is that I need to rank model variables by their importance and Weka doesn't have such an option. On the other hand the 'weight by tree importance' operator of RM expects to receive a RM based tree model to operate.
The reason that I'm interested to use Weka RF implementation is the way much better performance results it produces compared to the RM native RF operator on all of my datasets. It also seems to be impossible to mimic the Weka RF behaviour by tweaking RM RF parameteres.
Do you have any suggestions regarding that?
0 -
I also experienced the same problem with weka models in trees, they dont show up and the message persists..
0
Altair Employee