I have a list of different types of transactions per day. I can count the number of transactions per type per day and then create a boxplot over all days and even more I can create those boxplot by weekday. I use a filter to look for a particular transactions boxplots, day of week as dimension and then an expression to aggregate the counts. However, when the count of that transaction is zero for a date, the boxplot does not take into account that date so the fractiles are wrong. For instance, if transaction type A happend 2 times over one date out of 365, my boxplot looks like all the fractiles are 2 when they should be zero. Any ideas? thanks.
This is the expression I am using in the boxplot fractiles:
Fractile(aggr(count(transactionID),appt_date),0.25)