https://github.com/tpeterw/probabilistic-majority-voting

Generic Majority Voting procedure for probabilities of categories in ML classification algorithms (Available on PyPI)
https://github.com/tpeterw/probabilistic-majority-voting

machine-learning majority-judgement-voting pypi python python3

Last synced: about 1 month ago
JSON representation

Generic Majority Voting procedure for probabilities of categories in ML classification algorithms (Available on PyPI)

• Host: GitHub
• URL: https://github.com/tpeterw/probabilistic-majority-voting
• Owner: TPeterW
• License: mit
• Created: 2016-11-27T20:44:04.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2017-04-06T06:41:35.000Z (over 7 years ago)
• Last Synced: 2024-10-13T00:01:30.936Z (about 1 month ago)
• Topics: machine-learning, majority-judgement-voting, pypi, python, python3
• Language: Python
• Homepage: https://pypi.org/project/majorityvoting/
• Size: 13.7 KB
• Stars: 0
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Probabilistic Majority Voting

Generic Majority Judgement Voting procedure for probabilities of categories in machine learning




## Usage

To install through PyPI in command line

    pip install majorityvoting

And to use majority judgement voting in your code

    from majorityvoting import MajorityVoting

    voter = MajorityVoting.Voter(probabilities, categories)

    indices, names = voter.vote(windows=5)

The initializer takes in 2 arguments

* probabilities - a `numpy.ndarray` of lists (2d array) with equal length, each embedded list contains the probability of categories predicted by some machine learning algorithm, exactly the same you will get from [scikit-learn](http://scikit-learn.org/stable/)'s [`predict_proba()`](http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html#sklearn.linear_model.LogisticRegression.predict_proba) function

* categories - a list of names of categories, each category must match the index of it's probability in each and every embedded list in `probabilities`

## TODOs

1. It's a pain to manually check the size of the lists and embedded ones, will require input argument to be `pandas.DataFrame` type in the next version