Comments on: Calculating running variance in Python and C++

By: Tom MacWright

Tom MacWright — Wed, 27 Jun 2012 17:37:56 +0000

Thanks for posting this code! It was super-informative and a great shortcut to implementing standard deviation & variation in javascript with streams – https://github.com/tmcw/stream-statistics

By: brendano

brendano — Wed, 03 Dec 2008 17:55:29 +0000

Pretty cool, thanks!

Is windowed variance even harder?

By: Daniel Lemire

Daniel Lemire — Wed, 03 Dec 2008 17:51:49 +0000

I thought you were after an windowed variance.

If you are ever interested, you can compute the windowed max-min in constant time, irrespective of the window size! This is not entirely trivial, by the way. See my code:

http://code.google.com/p/maxminfilters/

By: brendano

brendano — Sun, 30 Nov 2008 20:51:06 +0000

could use a python for loop, but it’s slower than having numpy do an array operation because it’s in C :)

i guess figuring out how to push array operation into low-level libraries is a step simpler than matrix operations in low-level libraries…

By: Patrick

Patrick — Sun, 30 Nov 2008 08:06:50 +0000

Thanks for the plug but BLAS is pretty irrelevant here. Any decent array library will support a fold opration to give you the result with a constant amount of memory.

Also, in python, couldn’t you just use a for loop?