2D histograms

[yakir12]

This would be very powerful for processing coordinates, for instance:

xys = [[rand(2) for _ in 1:10] for _ in 1:3] # 3 tracks with 10 coordinates each
o = Hist(.2:0.1:.8, .3:.1:1) # dispatch on 2 arguments versus one, so no need for new name
for xy in xys # iterate over tracks
    fit!(o, xy) # fit all the coordinates
end

[joshday]

That exists in https://github.com/joshday/AverageShiftedHistograms.jl, but it would be nice to have it built into OnlineStats

[yakir12]

I guess a time will come when people will want to (on-line) fit n-dimensional histograms to their data... So it might be good to implement it in a generic enough way so to handle that as well. I have no idea what I'm talking about: maybe that is theoretically impossible.

[joshday]

I think StatsBase.Histogram can do N-dimensional fitting. If I remember correctly, there's also a low level function that does online updating.

Also, apparently I forgot about OnlineStats.HeatMap which does exactly what you want.

I think I like keeping HeatMap separate from Hist in order to keep the code cleaner (less type parameters, etc.)