The rapid deployment workspace node referenced in KB article that shows how to generate and deploy advanced I-tree analysis PMML model and deploy it., did not write back to the input data file when I checked the appropriate box within the "Save Results" area of the node. It seems to recognize the correct input data file when it gives me options for assigning the "variable list to write back prediction statistics to" because it gives me the list of variables from the correct file. However, the predictions don't end up in the file after the workspace is run. Any ideas on how to remedy this would be greatly appreciated.
OK, I see what I was missing now... When the Subset node is placed in the workspace, the prediction data is written into that Subset node which contains all the same variables as the original input file. Then if I open the Subset node, the predictions are all there so that I can copy and paste them into the original input file. That will work for me. Thanks!
The Rapid Deployment node requires a input data node that connect to Rapid Deployment node as input file for prediction. In that KB article, "Subset-TestingData" node connects to Rapid Deployment node and serves as input file for model deployment for prediction. The Raid Deployment can detect the input data node that connect to it, take input data and make prediction for that input data by deploying the model.
You can also load other data file with same variables names, and connect to Rapid deployment node, and make predictions.
I have created a KB article on how to save model deployment prediction result and write back to original input file.
You can also refer to attached workspace with examples to save and write prediction result back to input file.
In the KB article you created, I see that instead of using the Subset-Testing Data node as input to the Rapid Deployment node, you used another spreadsheet. Is that because the prediction data will not write back to the original input data if a Subset node is placed between the input data and the Rapid Deployment node?
What I would ultimately like to be able to do is select specific cases on which to deploy the C&RT model and have the prediction data written back to the original input file. Is this possible? Can the prediction data feed back past the Subset node to the original input file?
Glad that you have figured that out. Yes. You can still use "Subset-TestingData" node as input for Rapid Deployment node, in that case, you will need to create two new variables in the "Subset-TestingData" node spreadsheet and the prediction result will be written back to the "Subset-TestingData" node too. I used a separate spreadsheet as input simply for the illustration convenience purpose.