In rare occasions the Server seems to submit extremely large snipe per minute counts, that mess up the statistics page.

We should check for negative/very large numbers and
a) scale them down some how (wouldn't recommend this)
b) reset the count and submit 0 (that's what I'd do)

Well also, 3) Snipes per minute does not seem like a very useful stat
anyway. I would want to know blocks changed per minute, not snipes.
I.e., accounting for the size of the snipes as well.

On Tue, Oct 2, 2012 at 11:04 AM, Gavin Jenkins [email protected]:

1. That's not very surprising. Often, I will set my brush to size Zero,
   and just hold down the mouse button and roll my cursor around randomly,
   e.g. to rough up the surface of leaves in notch trees without messing up
   the middle. That would be hundreds of snipes per minute.

2. If you can export the statistics data to something like excel or
   whatever, then it would be trivial to simply exclude individual minutes
   where the number of snipes is above 2 standard deviations from the mean, or
   some such. Then analyze/plot the remainder of the data.

On Mon, Oct 1, 2012 at 7:27 PM, Mike Matrix [email protected]:

In rare occasions the Server seems to submit extremely large snipe per
minute counts, that mess up the statistics page.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/49.

1. That's not very surprising. Often, I will set my brush to size Zero,
   and just hold down the mouse button and roll my cursor around randomly,
   e.g. to rough up the surface of leaves in notch trees without messing up
   the middle. That would be hundreds of snipes per minute.

2. If you can export the statistics data to something like excel or
   whatever, then it would be trivial to simply exclude individual minutes
   where the number of snipes is above 2 standard deviations from the mean, or
   some such. Then analyze/plot the remainder of the data.

On Mon, Oct 1, 2012 at 7:27 PM, Mike Matrix [email protected]:

In rare occasions the Server seems to submit extremely large snipe per
minute counts, that mess up the statistics page.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/49.

Well, even if someone held down the mouse button this will not result in more than a billion+ snipes.
Also, exporting and manipulating is impossible since the data gets sent to the mcstats server(s) and the server will then parse and plot the data.
Anyways, block changes per minute seems pretty useful I'll crate an issue for that one.