Question
EY Global Delivery Services India LLP
HK
Last activity: 29 May 2018 7:16 EDT
Champion Challenger
I cannot find enough material on the champion challenger topic to understand its full working. The only material I could find is in the practice section in the Pega Decision Consultant. Can anyone help me in finding some relevant material on the champion challenger topic in pdn.
Regards
Aritra Ganguly
Accepted Solution
Pegasystems Inc.
NL
The champion challenger component basically just generates a random number and selects between its inputs based on that
It is typically used to test a model, strategy or treatment on a small, random, set of your population. Then after monitoring the results you change the percentages or redesign the propositions, strategies or models if they're not giving you the performance/success rate that you're after.
Pegasystems Inc.
IN
Hello Aritra,
Please go through below articles and check for champion challenger ,hope it helps
https://docs-previous.pega.com/extending-topcreditcardoffers-strategy-711
https://community.pega.com/sites/default/files/help_v74/express/procomhelpmain.htm
Thanks,
Arun
Pegasystems Inc.
IN
Hi AritraG1,
Please find the below link for reference and let me know if you have any specific queries on champion challenger
https://docs-previous.pega.com/dsm-getting-started-guide-714
EY Global Delivery Services India LLP
HK
Does the champion challenger switches the champion if a challenger has the potential to become a champion or do we have to change it manually.
Regards
Aritra
Accepted Solution
Pegasystems Inc.
NL
The champion challenger component basically just generates a random number and selects between its inputs based on that
It is typically used to test a model, strategy or treatment on a small, random, set of your population. Then after monitoring the results you change the percentages or redesign the propositions, strategies or models if they're not giving you the performance/success rate that you're after.
Pegasystems Inc.
NL
If you want to do something more 'automated' then you would use something like a priority component typically. Say as an example we have multiple cross sell offers for a customer, and adaptive models that calculate the likelihood of a response, a priority component can the output the top (n) offers with highest predicted response. Ofcourse you can throw other weightings into the mix to calculate priority, and also exclude offers (decision alternatives) based on hard exclusion rules.