Strategic Analytics and SAS

Deploying a Promoter Score Model on a Larger Customer Base

Even though the Gradient Boosting model is slightly better than the Regression model, we’re going to use the Regression model to score the entire data set with its predictions. This is so that we can understand which text topic(s) are key drivers for the Promoters, Passives, and Detractors. Once the entire customer data set is scored with the prediction, we can then aggregate our results and develop some high-level model results rather than detailed individual customer scores. Typically, one may apply this scoring model by deploying it in a batch process where the scoring code (SAS, Java, or C) can be executed at a day and time of one’s choosing at regularly scheduled intervals. ...