https://github.com/johannesbuchner/fasterscipydists

fasterscipydists provides faster scipy.stats distributions
https://github.com/johannesbuchner/fasterscipydists

probability-distribution scipy statistics

Last synced: 2 months ago
JSON representation

fasterscipydists provides faster scipy.stats distributions

• Host: GitHub
• URL: https://github.com/johannesbuchner/fasterscipydists
• Owner: JohannesBuchner
• Created: 2023-05-24T14:27:57.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-05-24T14:45:53.000Z (over 1 year ago)
• Last Synced: 2024-04-16T07:05:46.795Z (10 months ago)
• Topics: probability-distribution, scipy, statistics
• Language: Python
• Homepage:
• Size: 3.91 KB
• Stars: 5
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.rst

Awesome Lists containing this project

README

        
fasterscipydists

=================

fasterscipydists provides faster scipy.stats distributions.

Distributions implemented:

* uniform

* norm (Gaussian)

* expon (Exponential)

* dirac (Delta function)

Functions implemented:

* mean

* std

* ppf

* pdf

* logpdf

* cdf

Contributions are welcome! Please open issues or pull requests!

Usage

-----

This is a drop-in replacement, so instead of::

	import scipy.stats

	rv = scipy.stats.norm(1, 2)

you would do::

	import fasterscipydists

	rv = fasterscipydists.norm(1, 2)

See the scipy.stats documentation https://docs.scipy.org/doc/scipy/reference/stats.html

Speed-up

--------

speed.py reports these numbers::

	norm.pdf       :  7.6x faster with scipy.stats=0.126s, this=0.015s

	norm.logpdf    : 10.8x faster with scipy.stats=0.120s, this=0.010s

	norm.cdf       : 10.3x faster with scipy.stats=0.087s, this=0.008s

	norm.ppf       : 15.9x faster with scipy.stats=0.149s, this=0.009s

	uniform.pdf    :  9.2x faster with scipy.stats=0.114s, this=0.011s

	uniform.logpdf :  8.2x faster with scipy.stats=0.127s, this=0.014s

	uniform.cdf    :  2.4x faster with scipy.stats=0.084s, this=0.025s

	uniform.ppf    : 26.7x faster with scipy.stats=0.144s, this=0.005s

	expon.pdf      : 15.3x faster with scipy.stats=0.110s, this=0.007s

	expon.logpdf   : 26.7x faster with scipy.stats=0.112s, this=0.004s

	expon.cdf      : 10.9x faster with scipy.stats=0.097s, this=0.008s

	expon.ppf      : 21.8x faster with scipy.stats=0.154s, this=0.007s

Tests

-----

Systematic verification against scipy.stats is done in test.py. Run with::

	pytest test.py