Skip to main content
Join the official 2019 Python Developers SurveyStart the survey!

Badfish - A missing data analysis and wrangling library in Python

Project description

Badfish introduces MissFrame, a wrapper over pandas DataFrame, to wrangle through and investigate missing data. It opens an easy to use API to summarize and explore patterns in missingness.

Badfish provides methods which make it easy to investigate any systematic issues in data wrangling, surveys, ETL processes which can lead to missing data.

The API has been inspired by typical questions which arise when exploring missing data.

Badfish uses the where and how api in most of its methods to prepare a subset of the data to work on. where : Work on a subset of data where specified columns are missing. how : Either all | any of the columns should be missing.

Eg. mf.counts(columns = ['Age', 'Gender']) would give counts of missing values in the entire dataset.

While, mf.counts(where=['Income'], columns = ['Age', 'Gender']) would give counts of missing values in subset of data where Income is already missing.

Installation

pip install badfish

Usage

>>> import badfish as bf
>>> mf = bf.MissFrame(df)

Example

Will add an exmaple IPython notebook soon.

Counts

Basic counts of missing data per column.

>>> mf.counts(where=['gender', 'age'], how='all', columns=['Income', 'Marital Status'])

Pattern

Get counts on different combinations of columns with missing data. True means missing and False means present.

>>> mf.pattern()

The same can be visualized in the form of a plot (inspired by VIM package in R)

>>> mf.plot(kind='pattern')

Example plot:

Note: Both where and how can be used in this method.

Itemset Mining

Use frequency item set mining to find subgroups where data goes missing together. Note: This uses the PyMining package.

>>> itemsets, rules = mf.frequency_item_set()

Cohort

Tries to find signigicant group differences between values of columns other than the ones specified in the group clause. Group made on the basis of missing or non-missing of columns in the group clause. Internally uses scipy.stats.ttest_ind.

This method works on the values in each column instead of column names.

Note: Experimental method.

>>> mf.cohort(group=['gender'], columns=['Income'])

License

Please see the repository license.

Generally, we have licensed badfish to make it as widely usable as possible.

Call for contribution

If you have any ideas, issues or feature requests, feel free to open an issue, send a PR or contact us.

Project details


Download files

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

Files for badfish, version 0.1.2
Filename, size File type Python version Upload date Hashes
Filename, size badfish-0.1.2-py3.5.egg (15.5 kB) File type Egg Python version 3.5 Upload date Hashes View hashes
Filename, size badfish-0.1.2.zip (14.1 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page