February 18, 2009 at 12:54 pm
I need to compare a model built in SQL Server with a model from another data mining application. The only metric I have from the other application is a ROC curve and the area underneath it.
First, is there a way to get a ROC curve in SQL server?
Second, if not, how can the Lift score be converted to, or compared with, or normalized with, the area under the ROC curve?
thanks much!!
b
May 28, 2009 at 12:02 pm
The ROC curve area measures the probability that a random positive example has a higher assigned score than random negative example. ROC curves are similar to lift charts in that they provide a means of comparison between individual models and determine thresholds which yield a high proportion of positive hits. When the number of all potential variable being studied P is unknown, true positive TP rate cannot be computed and ROC curve cannot be used, the lift chart is used for measuring model accuracy.
In your situation, you might want to consider comparing True Positive TP rates.
Hope that helps.
Jen Underwood
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply