Priority scores/ propensity scores
Hi,
Right now I am looking for the priority scores/ propensity scores for the best offer , 2nd best offer, 3rd best offer etc. from adaptive models and planning to populate them in our Dashboard report with third party software. I wonder if someone can help us find the name of the data/table with such information in PEGA and where to locate the data/table. I only can find the scores for the best offer. Thank you for your help.
Pegasystems Inc.
US
Can you elaborate on your need please. The system outputs a unique result for every customer of their actions (offers). So there results are customer centric...not offer centric (meaning there is no top offer across the system, which looks like what you might be trying to display on a dashboard). If you can clarify what exactly you are trying to accomplish and why, that might allow others to give you some additional help.
ally
US
Yes, I am looking for the unique result for each customer by each offer in the next-best-action for arbitration. For example, there are 4 offers(treatments) available, I wonder if PEGA contains such table with propensity score for each customer by each offer,
|
|
Propensity of Offer1 |
Propensity of Offer2 |
Propensity of Offer3 |
Propensity of Offer4 |
|
Customer1 |
0.8 |
0.6 |
0.5 |
0.4 |
|
Customer2 |
0.4 |
0.5 |
0.3 |
0.7 |
What’s more I am also looking for table containing final prioritization value (Propensity (P)* Context weighting (C) * Action value (V) *Business levers (L)) for each customer by each offer which is used to decide which offer will be presented to the customer, something like
|
|
prioritization of Offer1 |
prioritization of Offer2 |
prioritization of Offer3 |
prioritization of Offer4 |
Yes, I am looking for the unique result for each customer by each offer in the next-best-action for arbitration. For example, there are 4 offers(treatments) available, I wonder if PEGA contains such table with propensity score for each customer by each offer,
|
|
Propensity of Offer1 |
Propensity of Offer2 |
Propensity of Offer3 |
Propensity of Offer4 |
|
Customer1 |
0.8 |
0.6 |
0.5 |
0.4 |
|
Customer2 |
0.4 |
0.5 |
0.3 |
0.7 |
What’s more I am also looking for table containing final prioritization value (Propensity (P)* Context weighting (C) * Action value (V) *Business levers (L)) for each customer by each offer which is used to decide which offer will be presented to the customer, something like
|
|
prioritization of Offer1 |
prioritization of Offer2 |
prioritization of Offer3 |
prioritization of Offer4 |
|
Customer1 |
0.8 |
0.6 |
0.5 |
0.4 |
|
Customer2 |
0.4 |
0.5 |
0.3 |
0.7 |
Once I have these propensity/ prioritization values, I will use them in the dashboard report. Thanks.
Pegasystems Inc.
US
Pega only captures such data of the offer was actually sent or presented to the customer. If a decision was made across 4 offers, but only 1 offer was sent/presented, then only 1 record with this data would be retained. This data is stored in interaction history in the fact table in the pyPropensity and pxPriority fields. If all 4 offers were sent/presented, then the fact table would store an pending or impression level record for each of these as a separate row for a given customer, so the structure is different than you have outlined. Its more like this:
| pySubjectID | pyName (offer) | pyPropensity | pxPriority | Other attributes... |
|---|---|---|---|---|
| Customer1 | Offer A | .8 | .85 | |
| Customer1 | Offer B | .6 | .66 | |
| Customer2 | Offer A | .4 | .44 | |
| Customer2 | Offer C | .3 | .33 | |
| Customer2 | Offer D | .7 | .77 |
Customer 1 was presented Offer A and B, customer 2 was presented Offer A, C and D.
Pega only captures such data of the offer was actually sent or presented to the customer. If a decision was made across 4 offers, but only 1 offer was sent/presented, then only 1 record with this data would be retained. This data is stored in interaction history in the fact table in the pyPropensity and pxPriority fields. If all 4 offers were sent/presented, then the fact table would store an pending or impression level record for each of these as a separate row for a given customer, so the structure is different than you have outlined. Its more like this:
| pySubjectID | pyName (offer) | pyPropensity | pxPriority | Other attributes... |
|---|---|---|---|---|
| Customer1 | Offer A | .8 | .85 | |
| Customer1 | Offer B | .6 | .66 | |
| Customer2 | Offer A | .4 | .44 | |
| Customer2 | Offer C | .3 | .33 | |
| Customer2 | Offer D | .7 | .77 |
Customer 1 was presented Offer A and B, customer 2 was presented Offer A, C and D.
Note: The Interaction history store is a star schema data store and it is an operational store that should not be reported on directly in production. All reporting should be done external to the production system from a periodic extract of this store. Information on the schema for this store can be found here: https://community.pega.com/knowledgebase/articles/decision-management-overview/interaction-history-data-model-pega-72-73
As a closing note, a dashboard report at customer level as you suggest above seems very odd. You would have millions of rows being output. I'm not sure of the value of such a report. I think you might want to consider what your true intent here should be. I've provided the data above just as a means to help you understand the data, but I would not recommend such a report.
ally
US
The pyPropensity/pxPriority values shown in the Interaction History Fact table are for final and best offer(s), meaning only these offers will be presented to the customers based on the highest calculated pyPropensity/pxPriority values. Are there any intermediate tables available before the interaction history fact table with pyPropensity/pxPriority values for all offers not just for the offer(s) presented to the customers in PEGA?