Zoom levels and feature aggregation

[ekastimo]

During the binning process (creating geo-cells), we compute the number of mobile money agents in the cell. Then obtain the population of the cell. These two are then used to generate People per mobile money agent Data i.e.
PeoplePerAgent = No of People in cell/ No of Mobile money agent
This value is then added as one of the properties of the bin(Geocell).
On the client side, we use dynamic map box style to allow a user to filter the data

{ 'id': 'mobilemoney-aggregated-0', 'type': 'fill', 'source': 'osm-mobilemoney-raw', 'source-layer': 'osm', 'maxzoom': 12.01, 'paint': { 'fill-color': '#ffe4b2' }, 'filter': ['all', ['>=', '_peoplePerAgent', 20], ['<', '_peoplePerAgent', 55] ] }
So this is a pictorial representation of the problem

We need to find an optimal way of fixing this
@awright @cgiovando @PaulUithol please advise on this

[krkwesiga]

@cgiovando could help us get some traction in terms of ideas for this challenges from the dev community