SAS Data Science

Building models with SAS Enterprise Miner, SAS Factory Miner, SAS Viya (Machine Learning), SAS Visual Text Analytics, with point-and-click interfaces or programming
BookmarkSubscribeRSS Feed
Ronein
Meteorite | Level 14

Hello,

I want to ask about a PD(probability of default) model in bank indusrty that provide estimated probability for default in next 12 months  for the investigated population.

In order to peform monthly monitoring validation of the model we are using 2 measurements:

1- Gini 

2-Percentge of catch till percentile 20.

I would like to ask about this measurement "Percentge of catch till percentile 20".

The step that I am using to calculate it are:

1-We calculate the predicted PD and then sort the customers by their PD from high to low.

2-Then we calculate the value of 20th percentage of PD.

3-Then we mark the customers who are until 20th percentage.

4-Then we count how many failures(Touch default) these customers(who were marked) have.

5-Then we count how many failures (Touch default)  there are for all population

6-Then we divide  (4)/(5) and this is the measurement of "Percentge of catch till percentile 20"

Model is "better" If "Percentge of catch till percentile 20" is higher

 

My questions:

I didn't find such measurement over internet search.

May anyone tell what is the name of this measurement and if it is familar?

I also want to ask If SAS can calculate this measurement (in one step)?

 

 

 

1 REPLY 1
sbxkoenk
SAS Super FREQ

Hello,

 

I have moved your message to the 'SAS Data Mining and Machine Learning' board.

 

What you ask for is all pre-calculated in Enterprise Miner (SAS 9.4) and Model Studio (SAS VIYA VDMML).

See here :

https://go.documentation.sas.com/doc/en/emref/15.2/p01jgc9rmzsg37n1lfncp67t0unm.htm

 

Look at all the measures below this phrase:
The following terminology is used for the Gains and Lift Chart:

 

Thanks,

Koen

sas-innovate-white.png

Join us for our biggest event of the year!

Four days of inspiring keynotes, product reveals, hands-on learning opportunities, deep-dive demos, and peer-led breakouts. Don't miss out, May 6-9, in Orlando, Florida.

 

View the full agenda.

Register now!

How to choose a machine learning algorithm

Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 733 views
  • 2 likes
  • 2 in conversation