Hi Karthik,
Max’ suggestion did not really help me for my case (but very handy).
In fact, I did not formulate correctly my question:
I want some calculated values from my data for this month and this week (reference date = now). Obviously, “week” values are a subset of “month” values (i.e. included in “month” value).
Max’s answer is calculations partitioned over a field, i.e. exclusive, the calculations I want are cumulative. My solution (not sure it’s the best one) give me the correct values I was looking for and is as follow.
1- Create a calculated field “metricDate_UTC_segment”:
ifelse(truncDate(‘MM’,{metricDate_UTC}) >= truncDate(‘MM’, ${Today}), 1, 0)
+
ifelse(truncDate(‘WK’,{metricDate_UTC}) >= truncDate(‘WK’, ${Today}), 1, 0)
Where 0 = out of date range, 1 = inside largest date range (month), 2 = inside 2nd date range (week).
2- New calculated fields using runningSum, runningAvg, ex.:
runningSum(distinct_count(conversationId), [{metricDate_UTC_segment} DESC])
runningAvg(avgIf(value, metric = ‘tHandle’), [{v0.0 metricDate_UTC_segment} DESC]) / 1000.0
3- New calculated field for getting date range instead of 0, 1, 2, Pivot table with:
- Rows: Date range, metricDate_UTC_segment (hidden, and level 0 hidden)
- Columns: my calculated fields (point 2)

Got what I wanted!
Now If I may, as you’re from QuickSight team, one comment / suggestion:
I obtained the same results at the beginning, very quickly, with simple functions like distinct_count, avgIf and adding 2 visual with different filters based on parameters. Cool, easy, quick, but ugly: impossible to merge the 2 visuals, swap rows/columns gives an undefined/non editable column header:

QS feature suggestion: adding filter for individual column (and row?) not just by visual.
Best,
Lionel