https://github.com/linsomniac/python-movingaverage
Function for computing a moving average of data.
https://github.com/linsomniac/python-movingaverage
Last synced: 7 months ago
JSON representation
Function for computing a moving average of data.
- Host: GitHub
- URL: https://github.com/linsomniac/python-movingaverage
- Owner: linsomniac
- Created: 2011-02-05T22:41:52.000Z (over 14 years ago)
- Default Branch: master
- Last Pushed: 2011-02-16T03:53:47.000Z (over 14 years ago)
- Last Synced: 2025-04-02T23:09:01.244Z (7 months ago)
- Language: Python
- Homepage:
- Size: 91.8 KB
- Stars: 20
- Watchers: 3
- Forks: 7
- Open Issues: 0
-
Metadata Files:
- Readme: README
Awesome Lists containing this project
README
Written by Sean Reifschneider
Released into the Public Domain
2011-02-05
Module to compute the moving average of a list. For example:
from movingaverage import movingaverage
print list(movingaverage([1,2,3,4,5,6], 3))
>>> [2,3,4,5]
NOTE: Another option is the "stats" pacakge, which is currently (Feb 2011)
in early development. The author of that package says that it includes
several moving average functions. I haven't used it, because I didn't find
out about it until after I wrote this. For more information see:
http://pypi.python.org/pypi/stats/
movingaverage(data, subset_size, data_is_list = None, avoid_fp_drift = True)
Return the moving averages of the data, with a window size of
`subset_size`. `subset_size` must be an integer greater than 0 and
less than the length of the input data, or a ValueError will be raised.
`data_is_list` can be used to tune the algorithm for list or iteratable
as an input. The default value, `None` will auto-detect this.
The algorithm used if `data` is a list is almost twice as fast as if
it is an iteratable.
`avoid_fp_drift`, if True (the default) sums every sub-set rather than
keeping a "rolling sum" (which may be subject to floating-point drift).
While more correct, it is also dramatically slower for subset sizes
much larger than 20.
NOTE: You really should consider setting `avoid_fp_drift = False` unless
you are dealing with very small numbers (say, far smaller than 0.00001)
or require extreme accuracy at the cost of execution time. For
`subset_size` < 20, the performance difference is very small.