Hi Kimmo,
The "scorecard" from the Reporter node is not a true scorecard like a scorecard that you would use in the credit scoring industry. The main difference is that you can interpret the scorecard points of a true scorecard as the probability of an event doubling. For example, the Scorecard node (available if you have a credit scoring license) has points to double odds set to 20 (you can change this property to another value). This means that someone with a score of 740 has double the odds of being a default compared to someone with a 720 score.
The scorecard from the Reporter node is great to help you understand what is driving your model. Whether you used a neural network, a regression, a random forest, or any other model, you can use the points to compare groups within each variable. Unfortunately, I don't think you can use this scorecard to score future data.
For example, you can use the Reporter node scorecard from your screenshot to notice that the event you are modeling is related to lower groups of age, or that it is very related to low first-member-begin-months and just a tiny bit related to high first-member-begin-months. But since you cannot compare the 45 points of low age with the 71 points of low first-member-begin-months, you cannot compare a 720 score with a 740 score.
If you really wanted to use this scorecard results, below an idea. But again, this scorecard is intended as a summary, not as a real tool to score future data.
Notice that the Reporter node creates a SAS data set called report_scorecard in your project workspace. For example if my project is called MiguelDataMining, my first workspace would be in MiguelDataMining\Workspaces\EMWS1. If you want the scorecard from your first Reporter node, browse to find report_scorecard. If you had two or more Reporter nodes, find report2_scorecard, report3_scorecard, etc. It shouldn't take you much effort to convert this dataset into scoring code.
I hope it helps!
FYI, for anyone not familiar with the Reporter node, this option is available when you specify the Nodes property as Summary. A SAS program called Rapid Predictive Modeler creates this summary output for you.
Thanks,
Miguel
