I am trying different ways of implementing statistic tests in my projects - at the dashboard level- not in the scripting.
I am mainly using the TTest - in particular the function ttest_sig( Group, Value ).
According to the manual the field Value is where this function calculates the mean (and the difference between the 2 Groups).
This works fine if you have metric data like e.g. volume, sales,...
What if you need to calculate the difference of frequencies e.g. Is the count of Product A in this group higher than in the other group? And what about Product B and C...? Has anybody succeeded in solving this using statements/set analysis?
I expected that I could use an aggr-function as the Value field ...something like aggr(Count(ID),Products,Groups)
Although QlikView seems to accept the aggr-function as Value field, nevertheless the result is incorrect.
I found a way to workaround in scripting a crosstable, where I count a case/person as 1 if he has the product and 0 if not. As the crosstable offers me an attribute-field (Products) and data-field - I can simply use the datafield as the Value and the attribute-field as my table-dimension. But in fact this would mean a lot of extra-scripting if I always I have to script extra-crosstables for each single field.
Would be interested in hearing about your experiences and suggestions.