Graph on the problem page should match the calculation on the front page

[16:23:49] <ev> mpt: bdmurray and I were talking at the client sprint about how the graph on the bucket page has wild data and seemingly not useful in its current form. I think this is partly because I've made a mistake and our calculation is:
[16:23:49] <ev> total instances of the problem across all releases / the number of unique systems seen in a 90 day period for a specific release
[16:23:49] <ev> I can fix not including the release in the numerator (by collecting more data), but in thinking about it more, I wonder if the ideal calculation would be the weighted instances of just this problem for a specific release divided by the unique systems seen in a 90 day period for the specific release.
[16:23:49] <ev> It's computationally more complex, but if that's going to give the best results then we can always try for it.[16:48:11] <mpt> ev, if I understand you correctly, the formula for the bucket graph is different from the formula for the front page graph when it should be the same.
[16:48:56] <ev> mpt: same with a smaller numerator for the bucket graph, yes
[16:49:47] <mpt> yes, and correspondingly smaller scale