Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Visual Data Mining and Machine Learning or just with programming

WOE in miner 7

Reply
Occasional Learner FO
Occasional Learner
Posts: 1

WOE in miner 7

Hi all!

 

My first question is about WOE transformations.

I am using interactive grouping node and miner by default assigns positive WOE values to below average bad rate and negative WOE values to above average bad rate. I would like to reverse the signs, where is the button with this option?

 

Thank you very much in advance for your help!

 

Super Contributor
Posts: 337

Re: WOE in miner 7

[ Edited ]

Hi FO,

Weight of evidence is different from the bad rate.

Considering that an event is an applicaiton flagged as credit default or "bad", and a non-event is a non-default or "good" application, we have that:

 

WOE is the log transformation of the percentage distribution of events to non-events, or in other words the log(event_i%/non-events_i%), where the ith % stands for the percentage of events or nonevents in that bin, relative to the total number of events or non-events.

Bad rate or event rate is the rate calculated as events/(events+non_events).

 

For certain cases you might want to reverse the woe lines by changing the order of your binary target from descending to ascending, but just make sure that your scorecard makes sense in terms of the higher the score, the lower the odds of an event happening. Another setting that might come handy is the option for reverse scorecard in the properties for the Scorecard node.

 

I hope this helps!

-Miguel

Ask a Question
Discussion stats
  • 1 reply
  • 285 views
  • 0 likes
  • 2 in conversation