Rolling Average Calculation

A Beast Mode is calculated on either a single record or group of records, depending on how the data on the card is being grouped. Because of that, the calculation doesn't have a way of reference other records. For example, I can't say, "I'm looking for the April average. Look back at other records that come before April and sum them up for me." You could create a card that does a quarterly average, though. To do that, set your date grain to "by Quarter" at the top of the card, then add your date and value columns to the X and Y axis, respectively. Then set your aggregation option for your value column to AVG.

Alternatively, you'd need to do the calcuation you're looking for in a Dataflow.  The transform would look something like this:

SELECT
  `DateField`
  ,`ValueField`
  ,(SELECT
      SUM(`ValueField`) / COUNT(*)  
    FROM
      TABLE t2
    WHERE
      t2.`DateField` > LAST_DAY(DATE_SUB(t1.`DateField`, INTERVAL 3 MONTH))
      AND t2.`DateField` < LAST_DAY(t1.`DateField`) ) AS `Rolling_Average`

FROM
  TABLE t1

This transform assumes that the DateField contains a date representing the month, e.g. "2015-01-01" for January.

We're using a subquery to go grab the sum of the last 3 months worth of values and then dividing them by the number of records it found in that range and calling that the Rolling_Average field. You'd need to replace your field names for `DateField` and `ValueField` (and add any others you want in the resulting dataset) and replace TABLE with the dataset table you're using as your input.

I hope that helps!