Skip to main content

Experiment Analysis Library

Project description

ExpAn: Experiment Analysis

A/B tests, or randomized controlled experiments, have been widely applied in different industries to optimize the business process and the user experience. ExpAn, or Experiment Analysis, is a Python library developed for the advanced statistical analysis of A/B tests.

The data structures and functions here are generic such that they can be used by both data scientists optimizing a user interface and biologists running wet-lab experiments. The library is also standalone and can be imported and used from within other projects and from the command line.

Major statistical functionalities include:

  • feature check

  • delta

  • subgroup analysis

  • trend

Table of Contents

Quick start

Install

To install simply run

python2 setup.py build
python2 setup.py install

And to test run:

python2 setup.py test

Some mock-up data

from expan.experiment import Experiment
from tests.test_data import generate_random_data

exp = Experiment('B', *generate_random_data())
exp.delta()

Further documentation

ExpAn Description - details about the concept of the library and data structures.

How to contribute

Style guide

We follow PEP8 standards with the following exceptions:

  • Use tabs instead of spaces - this allows all individuals to have visual depth of indentation they prefer, without changing the source code at all, and it is simply smaller

Branching / Release

We currently use the gitflow workflow. Feature branches are created from and merged back to the dev branch, and the master branch stores snapshots/releases of the dev branch.

See also the much simpler github flow here

Versioning

For the sake of reproducibility, always be sure to work with a release when doing the analysis!

We use semantic versioning (http://semver.org), and the current version of ExpAn is: v0.2.3.

The version is maintained in setup.cfg, and propagated from there to various files by the bumpversion program. The most important propagation destination is in version.py where it is held in the string __version__ with the form:

'{major}.{minor}.{patch}'

The __version__ string and a version() function is imported by core.__init__ and so is accessible to imported functions in expan.

The version(format_str) function generates version strings of any form. It can use git’s commit count and revision number to generate a long version string which may be useful for pip versioning? Examples: NB: caution using this… it won’t work if not in the original git repository.

>>> import core.binning
>>> core.version()
'v0.2.3'
>>> core.version('{major}.{minor}..{commits}')
'0.0..176'
>>> core.version('{commit}')
'a24730a42a4b5ae01bbdb05f6556dedd453c1767'

See: StackExchange 151558

Bumping Version

Can use bumpversion to maintain the __version__ in version.py:

$ bumpversion patch
or
$ bumpversion minor

This will update the version number, create a new tag in git, and commit the changes with a standard commit message.

When you have done this, you must push the commit and new tag to the repository with:

$ git push --tags

TODO

  • parallelization, eg. for the bootstrapping code

  • Bayesian updating/early stopping

  • multiple comparison correction, definitely relevant for delta and SGA, have to think about how to correct for time dependency in the trend analysis

  • implement from_json and to_json methods in the Binning class, in order to convert the Python object to a json format for persisting in the Results metadata and reloading from a script

License

The MIT License (MIT)

Copyright © [2016] Zalando SE, https://tech.zalando.com

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

History

0.2.0 (2016-05-03)

  • First opensource release to GitHub

0.1.0 (2016-04-29)

  • Relatively stable version used by Zalando internally, prior to opensource release

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

expan-0.2.3.tar.gz (66.1 kB view details)

Uploaded Source

Built Distribution

expan-0.2.3-py2.py3-none-any.whl (72.2 kB view details)

Uploaded Python 2 Python 3

File details

Details for the file expan-0.2.3.tar.gz.

File metadata

  • Download URL: expan-0.2.3.tar.gz
  • Upload date:
  • Size: 66.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No

File hashes

Hashes for expan-0.2.3.tar.gz
Algorithm Hash digest
SHA256 86f2f8af82646a92df2f27184b541eeedf2586317956d609f91b5faa39fc83cf
MD5 4d61e95093c84972d25e3217573ded4a
BLAKE2b-256 1ffbaac183ae0fb9cc91a9a3c005c967005394e2f8a3652ae754ebbdb5d18105

See more details on using hashes here.

File details

Details for the file expan-0.2.3-py2.py3-none-any.whl.

File metadata

File hashes

Hashes for expan-0.2.3-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 54cddf97ad71a6216e0c61ce813780722a0b577ff233cc73de43e31d53cbcbdd
MD5 08322774c0ab4cc7e36cbddcdfb5f0bd
BLAKE2b-256 7fe557d79e5ee2e05d3de11142b100cbf10ff794fedbaa6775565af9ede8959d

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page