Subscribe to DSC Newsletter

Hi,

Someone asked me a question on credit scoring model using WOE/IV only (without using logistic/OLS regression/CHAID).
 
Is it possible to give points to a variable based on its IV value and to the individual categories of that variable based on their WOE.
Suppose, the following is the output from the IV table, how would you use the WOE to give scores to them.
 
Asssumption - Total score is 100, significant vars - 4, Age is the most significant with IV - 0.46, other 3 vars have a lesser value than 0.46 but more than 0.25.
I thought we could weight a variable based on its IV value. Here Age's weight or score could be 100*0.46/(sum of IV of all 4 vars), which say, comes to 35.
 
Like this we can get IV weighted scores for each variable. But all the categories within a variable can't be assigned same weight. So, how to weight a variable's score within its categories. The best category can be given score equal to that of the variable and others be given scores lesser than that.
 
Do you think this is a correct approach and if so, how can get the scores for each category, 1,2,3,4,5 for age-group below ? Say - Group 5 as 35, 4 as 25, 3 as 20 and so on.
Age group           Good             Bad           Total  %Good %Bad %Good-%Bad Ln(%Good/%Bad) Marginal IV
1 50 120 170 8.00% 25.50% -17.50% -1.1604   0.20308535
2 75 110 185 12.00% 23.40% -11.40% -0.6680   0.07615328
3 125 90 215 20.00% 19.10% 0.90% 0.04348   0.00039137
4 175 80 255 28.00% 17.00% 11.00% 0.49774  0.05475144
5 200 70 270 32.00% 14.90% 17.10% 0.76480  0.13078134
                 
              IV 0.4651628

Thanks,

Nitin

Tags: EVIDENCE, INFORMATION, IV, OF, REGRESSION, Risk, VALUE, WEIGHT, WOE, scorecard

Views: 1242

Reply to This

On Data Science Central

© 2019   AnalyticBridge.com is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service