operdeck
commented
11 months ago 0 = everyone getting the same thing, no entropy 100 = everyone getting something unique
Everyone unique seems unlikely but if there is a top 3 with just 200 actions there are enough combinations to cover a very large customer base.
Op wo 20 sep. 2023 15:11 schreef yusufuyanik1 @.***>:
Is your feature request related to a problem?
We know that optionality matters for 1:1 experience. We can see how many actions we have in the action catalog and have an idea about how often we present them from the sizes of bubble. But there is no direct metric for optionality. Perhaps having 1 number that we can highlight if it is very low, it could be a good alarm. Describe the solution you'd like
The idea emerged from the plot below which shows that a customer keeps showing 5 actions over 80% of the time. Instead of showing a graph we can create a standard metric and interpret it cross-customers. [image: Screenshot 2023-09-20 at 14 59 31] https://user-images.githubusercontent.com/54443450/269291929-872d37b8-efcf-42ad-8733-6f6e95af9a51.png Describe alternatives you've considered
We can think about how to generalize the metric Additional context
No response
— Reply to this email directly, view it on GitHub https://github.com/pegasystems/pega-datascientist-tools/issues/120, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABEVTKCYZRQDA4QYUKBO7A3X3LTQRANCNFSM6AAAAAA4745XEY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
yusufuyanik1
Is your feature request related to a problem?
We know that optionality matters for 1:1 experience. We can see how many actions we have in the action catalog and have an idea about how often we present them from the sizes of bubble. But there is no direct metric for optionality. Perhaps having 1 number that we can highlight if it is very low, it could be a good alarm.
Describe the solution you'd like
The idea emerged from the plot below which shows that a customer keeps showing 5 actions over 80% of the time. Instead of showing a graph we can create a standard metric and interpret it cross-customers.
Describe alternatives you've considered
We can think about how to generalize the metric
Additional context
No response