tidypandas

A grammar of data manipulation for pandas inspired by tidyverse

tidypandas python package provides minimal, pythonic API for common data manipulation tasks:

• tidyframe class (wrapper over pandas dataframe) provides a dataframe with simplified index structure (no more resetting indexes and multi indexes)
• Consistent ‘verbs’ (select, arrange, distinct, …) as methods to tidyframe class which mostly return a tidyframe
• Unified interface for summarizing (aggregation) and mutate (assign) operations across groups
• Utilites for pandas dataframes and series
• Uses simple python data structures, No esoteric classes, No pipes, No Non-standard evaluation
• No copy data conversion between tidyframe and pandas dataframes
• An accessor to apply tidyframe verbs to simple pandas datarames
• …

Example

• tidypandas code:

df.filter(lambda x: x['col_1'] > x['col_1'].mean(), by = 'col_2')

• equivalent pandas code:

(df.groupby('col2')
   .apply(lambda x: x.loc[x['col_1'] > x['col_1'].mean(), :])
   .reset_index(drop = True)
   )

Why use tidypandas

tidypandas is for you if:

• you frequently write data manipulation code using pandas
• you prefer to have stay in pandas ecosystem (see accessor)
• you prefer to remember a limited set of methods
• you do not want to write (or be surprised by) reset_index, rename_axis often
• you prefer writing free flowing, expressive code in dplyr style

tidypandas relies on the amazing pandas library and offers a consistent API with a different philosophy.

Presentation

Learn more about tidypandas (presentation)

Installation

1. Install release version from Pypi using pip:

   pip install tidypandas
   

2. For offline installation, use whl/tar file from the releases page on github.

Contribution/bug fixes/Issues:

1. Open an issue/suggestion/bugfix on the github issues page.

2. Use the master branch from github repo to submit your PR.

---