a-ozbek / impyute

Data imputations library to preprocess datasets with missing data

Impyute

Impyute is a library of missing data imputation algorithms written in Python 3. This library was designed to be super lightweight, here's a sneak peak at what impyute can do.

>>> from impyute.datasets import random_uniform
>>> raw_data = random_uniform(shape=(5, 5), missingness="mcar", th=0.2)
>>> print(raw_data)
[[  1.   0.   4.   0.   1.]
 [  1.  nan   6.   4.  nan]
 [  5.   0.  nan   1.   3.]
 [  2.   1.   5.   4.   6.]
 [  2.   1.   0.   0.   6.]]
>>> from impyute.imputations.cs import mean_imputation
>>> complete_data = random_imputation(raw_data)
>>> print(complete_data)
[[ 1.    0.    4.    0.    1.  ]
 [ 1.    0.5   6.    4.    4.  ]
 [ 5.    0.    3.75  1.    3.  ]
 [ 2.    1.    5.    4.    6.  ]
 [ 2.    1.    0.    0.    6.  ]]

Feature Support

• Imputation of Cross Sectional Data

  • Multivariate Imputation by Chained Equations
  • Expectation Maximization
  • Mean Imputation
  • Mode Imputation
  • Median Imputation
  • Random Imputation

• Imputation of Time Series Data

  • Last Observation Carried Forward
  • Autoregressive Integrated Moving Average (WIP)
  • Expectation Maximization with the Kalman Filter (WIP)

• Dataset Generation

  • Datasets

    • MNIST
    • Random uniform distribution
    • Random gaussian distribution

  • Missingness Corruptors

    • MCAR
    • MAR (WIP)
    • MNAR (WIP)

• Diagnostic Tools

  • Loggers
  • Distribution of Null Values
  • Comparison of imputations
  • Little's MCAR Test (WIP)

Installation

To install impyute, run the following:

$ pip install impyute

Or to get the most current version:

$ git clone https://github.com/eltonlaw/impyute
$ cd impyute
$ python setup.py install

Documentation

Documentation is available here: http://impyute.readthedocs.io/

How to Contribute

Check out CONTRIBUTING