Some graphs spike for the last date

[digitarald]

A lot of graphs jump up or down for the last date:

The analysis should exclude the last x days for which telemetry doesn't provide a stable signal.

[pdolanjski]

This is likely the exclusion of XP/Vista users who moved to ESR. We should really include ESR in the overall numbers.

[pdolanjski]

One other option would be to not include ESR users (since the enterprise aspect will have a particular skew), but have an option to see the graphs with those numbers included.

[mreid-moz]

It would make sense that this was due to the migration of those users to ESR. March 5th is the beginning of the weekly report period per the analysis notebook, so it would include a couple of days during the migration period, which was apparent by March 9th. This would move many XP/Vista users off the release channel and hence out of this data set.

[digitarald]

Looks like know why this happened. Maybe we can clarify in the page?

[openjck]

Related: #28

In the longer-term, we could build the feature described in #28 to annotate specific points in charts

[openjck]

Closed with ac01c62

[Forkest]

OK, so we won't see ESR included. Where can I find the real number of XP/Vista users then?