periodToDatePercentile
The periodToDatePercentile
function calculates the percentile based on the
actual numbers in measure for a given time granularity (for instance, a quarter)
up to a point in time, relative to that period. It uses the grouping and sorting that are applied in the
field wells.
To return the nearest percentile value that is present in your dataset, use periodToDatePercentile
. To return an exact percentile value that might not be present in your dataset, use periodToDatePercentileCont
instead.
Syntax
periodToDatePercentile( measure, percentile, dateTime, period, endDate (optional))
Arguments
 measure

The argument must be a field. Null values are omitted from the results. Literal values don't work.
 percentile

The percentile value can be any numeric constant 0100. A percentile of 50 computes the median value of the measure.
 dateTime

The Date dimension over which you're computing PeriodToDate aggregations.
 period

The time period across which you're computing the computation. Granularity of
YEAR
meansYearToDate
computation,Quarter
meansQuarterToDate
, and so on. Valid granularities includeYEAR
,QUARTER
,MONTH
,WEEK
,DAY
,HOUR
,MINUTE
, andSECONDS
.  endDate

(Optional) The date dimension that you're ending computing periodToDate aggregations. It defaults to
now()
if omitted.
Example
The following example calculates the weektodate, 90th percentile of fare amount per payment type for the week of 063021. For simplicity in the example, we filtered out only a single payment. 063021 is Wednesday. QuickSight begins the week on Sundays. In our example. that is 062721.
periodToDatePercentile(fare_amount, 90, pickupDatetime, WEEK, parseDate("06302021", "MMddyyyy")