I have been asked by quite a few of our clients and I think this feature would be a very powerful addition to the views in DataExplorer.
The simplest use case is as follows:
I have a Bar/Line view with measure X and split by one single field [Field1]. With this new feature, I'd be able to say, "Show only top 10 entries" and an automatic filter will be attached to that Bar/LIne view.
Building on the simplest case, we can think of further generalisations like:
1. Configurable number of entries, i.e. 'top/bottom 'N' entries', N to be chosen by user 2. Top/Bottom 'N' entries, rather than only top entries 3. Extension of the same concept over aggregated measures 4. Function support to enable the same feature through formulae
We've experimented with a "Pareto" option in the Bar view in the past, and it's still there, but marked experimental because we might replace it with a better solution in future.
Your post above is a good starting point for discussing this.
Meanwhile if you want to explore the behaviour in the Bar view, see below. Don't rely on this in future version, and it may not work correctly in some situations:
- enable experimental features in the Alpha version - add a bar/line view, and reset it - stack by a number field - in the stack menu, change 'histogram' to 'pareto' - explore the further options in the stack menu.
These don't give you "top n", but rather they break the data into (e.g.) 5 "pots" of top 20%, 2nd 20%, etc. You can choose whether to measure pot size by number of records (e.g. producing 200 records per interval, if there are 1000 in total, or as near as possible) or by total field value (e.g. producing intervals each summing 200, if total for entire field is 1000, or as near as possible).