0

I'm looking for recommendations on a best practice here.

I have a requirement where on a given day I must have an arbitrary number of intervals (think buckets of time which are composed of transactions) where I can have at most N intervals per day. These intervals are like time but can be arbitrary lengths i.e. some are seconds, others are minutes.

How the intervals should be formed is based on my source data. On any given day, we always start with interval 1 and it is unknown the total number of intervals we will have by EOD, each interval is defined by a fixed number of transactions. For every interval I am going to need to know the end time as well.

What is the best approach here? Should I be bucketing my fact table and connecting to a standard hour/minute/second dimension or should I be using my transactional data to be making a dimension that accommodates it?

I appreciate your feedback.

Dave
  • 2,386
  • 1
  • 20
  • 38

1 Answers1

0

If the buckets are on time, you probably have to do it on one of your dimensions. There is a property on the attributes called bucket that can do that for you

Diego
  • 34,802
  • 21
  • 91
  • 134