Vitali Kaiser on 06 May 2016 16:01:03
Problem: If you working with large datasets you get often the "Too many values" warning and a wrong visual data representation, e.g., you plot a value over a timeline of a year with a sampling rate of seconds.
By applying a separate timeline filter you want to see a correct visual representation for time periods varying from several months to some seconds.
Solution: You should allow to group axis values. It would be enough if you could see let´s say 500 values on the timeline axis, e.g. if the there are values for a year you would only need 500 groups and therefore values. A single group would have to size of / 500.
The grouping and value calculation should be done in the (in-memory) database. That would also work quite good for DirectQuery, where you have this big data sets.
Would be nice if there would be several options for axis grouping, e.g. spatial grouping for Cartesian coordinates, or different bucket types for numbers (fixed sizes buckets or buckets on count of values).