03-30-2017 06:45 AM
Need help on deployment of GBM model in SAS Eminer.
We have developed the model using gradient boosting node in SAS Eminer. As it generates long scoring code, it gets truncated and provide path where the full code is saved. To check the model lift and do OOT, we have used the full score code using the following step:
filename score1 '/sasdata/SAS_MINING/EminerProjects/MH/Workspaces/EMWS1/Boost/EMPUBLISHSCORE.sas ';
GBM model provided higher lift compared to neural and logistic model. Formodel deployment, we execute the score node and score code export node. After executing the score code export node, we share the file details with the deployment team. unlike logistic and Neural model, deployment team is not able to deploy GBM model. They are getting the error while executing xml and scoring code. What can be a reason for this? Is it due to truncation of code even after running score code export node? Please help to resolve this issue. We want to figure out a way to get the complete code after running score code export node.
Details shared with deployment team (After running score code export node)
Folder Created: /sasdata/SAS_MINING/EminerProjects/MH/Score/sassrv_6
SAS Code: score.sas
Code XML: score.xml
Output XML: score_62.xml
Training Data: traindata.sas7bdat
Sample Data: scoredata.sas7bdat