Difference in between period and interval in DFE


#1

What is the difference in between period & interval in DFEs used in analytics


#2

period is the total length of time for which data will be retrieved. interval is the granularity of that data. E.g. if period = 30d, interval=day, then a time series will be input to your process function containing 30 days worth of data at day granularity.

How often an analytic gets triggered is turned off by default (marked by frequency annotation) and an analytics will get invoked every time data change happens. Users can choose to store their own state about when the analytics were invoked in types and then execute the business logic


#3
  1. Iā€™m not sure I understand why period should be different than interval. Most of the time they will be the same. Do you have a use case where period is different than interval?
  2. Can we still use the frequency annotation?

#4

Imagine you want all 365 daily values of a feature over the past year. In this case interval would be DAY and period would be ā€œ365d.ā€


#5

Thanks @rileysiebel

To answer 2, the frequency annotation is not available to be used right now. As mentioned above, you can chose not to do anything in the process function.