Skip to main content

Easy pipelines for pandas.

Project description

pdpipe
#########
|PyPI-Status| |PyPI-Versions| |Build-Status| |Codecov| |LICENCE|

Easy pipelines for pandas DataFrames.

.. code-block:: python

>>> df = pd.DataFrame(
data=[[4, 165, 'USA'], [2, 180, 'UK'], [2, 170, 'Greece']],
index=['Dana', 'Jack', 'Nick'],
columns=['Medals', 'Height', 'Born']
)
>>> pipeline = pdp.Coldrop('Medals').Binarize('Born')
>>> pipline(df)
Height Born_UK Born_USA
Dana 165 0 1
Jack 180 1 0
Nick 170 0 0

.. contents::

.. section-numbering::

Installation
============

Install ``pdpipe`` with:

.. code-block:: bash

pip install pdpipe

Some pipeline stages require ``scikit-learn``; they will simply not be loaded if ``scikit-learn`` is not found on the system, and ``pdpipe`` will issue a warning. To use them you must also `install scikit-learn`_.

.. _`install scikit-learn`: http://scikit-learn.org/stable/install.html


Features
========

* Pure Python.
* Compatible with Python 3.5+.
* A simple interface.
* Informative prints and errros on pipeline application.
* Chainning pipeline stages constructor calls for easy, one-liners pipelines.
* Pipeline arithmetics.


Design Decisions
----------------

* **Data science-oriented naming** (rather than statistics).
* **A functional approach:** Pipelines never change input DataFrames. Nothing is done "in place".
* **Opinionated operations:** Help novices avoid mistake by default appliance of good practices; e.g., binarizing (creating dummy variables) a colum will drop one of the resulting columns by default, to avoid `the dummy variable trap`_ (perfect `multicollinearity`_).
* **Machine learning-oriented:** The target use case is transforming tabular data into a vectorized dataset on which a machine learning model will be trained; e.g., column transformations will drop the source columns to avoid strong linear dependence.

.. _`the dummy variable trap`: http://www.algosome.com/articles/dummy-variable-trap-regression.html
.. _`multicollinearity`: https://en.wikipedia.org/wiki/Multicollinearity


Use
===

Pipeline Stages
---------------

Creating Pipline Stages
~~~~~~~~~~~~~~~~~~~~~~~

You can create stages with the following syntax:

.. code-block:: python

import pdpipde as pdp
drop_name = pdp.ColDrop("Name")


All pipeline stages have a predefined precondition function that returns True for dataframes to which the stage can be applied. By default, pipeline stages raise an exception if a DataFrame not meeting
their precondition is piped through. This behaviour can be set per-stage by assigning ``exraise`` with a bool in the constructor call. If ``exraise`` is set to ``False`` the input DataFrame is instead returned without change:

.. code-block:: python

drop_name = pdp.ColDrop("Name", exraise=False)


Applying Pipelines Stages
~~~~~~~~~~~~~~~~~~~~~~~~~

You can apply a pipeline stage to a DataFrame using its ``apply`` method:

.. code-block:: python

res_df = pdp.ColDrop("Name").apply(df)

Pipeline stages are also callables, making the following syntax equivalent:

.. code-block:: python

drop_name = pdp.ColDrop("Name")
res_df = drop_name(df)

The initialized exception behaviour of a pipeline stage can be overriden on a per-application basis:

.. code-block:: python

drop_name = pdp.ColDrop("Name", exraise=False)
res_df = drop_name(df, exraise=True)

Additionaly, to have an explaination message print after the precondition is checked but before the application of the pipeline stage, pass ``verbose=True``:

.. code-block:: python

res_df = drop_name(df, verbose=True)


Extending PipelineStage
~~~~~~~~~~~~~~~~~~~~~~~

To use other stages than the built-in ones (see `Types of Pipeline Stages`_) you can extend the ``PipelineStage`` class. The constructor must pass the ``PipelineStage`` constructor the ``exmsg``, ``appmsg`` and ``desc`` keyword arguments to set the exception message, application message and description for the pipeline stage, respectively. Additionally, the ``_prec`` and ``_op`` abstract methods must be implemented to define the precondition and the effect of the new pipeline stage, respectively.


Ad-Hoc Pipeline Stages
~~~~~~~~~~~~~~~~~~~~~~

To create a custom pipeline stage without creating a proper new class, you can instantiate the ``AdHocStage`` class which takes a function in its ``op`` contructor parameter to define the stage's operation, and the optional ``prec`` paramter to define a precondition (an alway-true function is the default).


Pipelines
---------

Creating Pipelines
~~~~~~~~~~~~~~~~~

Pipelines can be created by supplying a list of pipeline stages:

.. code-block:: python

pipeline = pdp.Pipeline([pdp.ColDrop("Name"), pdp.Binarize("Label")])


Pipeline Arithmetics
~~~~~~~~~~~~~~~~~~~~

Alternatively, you can add pipeline stages together:

.. code-block:: python

pipeline = pdp.ColDrop("Name") + pdp.Binarize("Label")

Or even by adding pipelines together or pipelines to pipeline stages:

.. code-block:: python

pipeline = pdp.ColDrop("Name") + pdp.Binarize("Label")
pipeline += pdp.ApplyToRows("Job", {"Part": True, "Full":True, "No": False})
pipeline += pdp.Pipeline([pdp.ColRename({"Job": "Employed"})])


Pipeline Chaining
~~~~~~~~~~~~~~~~~

Pipline stages can also be chained to other stages to create pipelines:

.. code-block:: python

pipeline = pdp.ColDrop("Name").Binarize("Label").ValDrop([-1], "Children")


Pipeline Slicing
~~~~~~~~~~~~~~~~

Pipelines are Python Sequence objects, and as such can be sliced using Python's slicing notation, just like lists:

.. code-block:: python

pipeline = pdp.ColDrop("Name").Binarize("Label").ValDrop([-1], "Children").ApplyByCols("height", math.ceil)
result_df = pipeline[1:2](df)


Applying Pipelines
~~~~~~~~~~~~~~~~~~

Pipelines are pipeline stages themselves, and can be applied to a DataFrame using the same syntax, applying each of the stages making them up, in order:

.. code-block:: python

pipeline = pdp.ColDrop("Name") + pdp.Binarize("Label")
res_df = pipeline(df)


Assigning the ``exraise`` paramter to a pipeline apply call with a bool sets or unsets exception raising on failed preconditions for all contained stages:

.. code-block:: python

pipeline = pdp.ColDrop("Name") + pdp.Binarize("Label")
res_df = pipeline.apply(df, exraise=False)


Additionally, passing ``verbose=True`` to a pipeline apply call will apply all pipeline stages verbosely:

.. code-block:: python

res_df = pipeline.apply(df, verbose=True)


Types of Pipeline Stages
========================

Basic Stages
------------

* AdHocStage - Define custom pipeline stages on the fly.
* ColDrop - Drop columns by name.
* ValDrop - Drop rows by by their value in specific or all columns.
* ValKeep - Keep rows by by their value in specific or all columns.
* ColRename - Rename columns.

Column Generation
-----------------

* Bin - Convert a continous valued column to categoric data using binning.
* Binarize - Convert a categorical column to the several binary columns corresponding to it.
* ApplyToRows - Generate columns by applying a function to each row.
* ApplyByCols - Generate columns by applying an element-wise function to columns.

Scikit-learn-dependent Stages
-----------------------------

* Encode - Encode a categorical column to corresponding number values.


Contributing
============

Package author and current maintainer is Shay Palachy (shay.palachy@gmail.com); You are more than welcome to approach him for help. Contributions are very welcomed, especially since this package is very much in its infancy and many other pipeline stages can be added.

Installing for development
--------------------------

Clone:

.. code-block:: bash

git clone git@github.com:shaypal5/pdpipe.git


Install in development mode with test dependencies:

.. code-block:: bash

cd pdpipe
pip install -e ".[test]"


Running the tests
-----------------

To run the tests, use:

.. code-block:: bash

python -m pytest --cov=pdpipe


Adding documentation
--------------------

This project is documented using the `numpy docstring conventions`_, which were chosen as they are perhaps the most widely-spread conventions that are both supported by common tools such as Sphinx and result in human-readable docstrings (in my personal opinion, of course). When documenting code you add to this project, please follow `these conventions`_.

.. _`numpy docstring conventions`: https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt
.. _`these conventions`: https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt


Credits
=======
Created by Shay Palachy (shay.palachy@gmail.com).


.. |PyPI-Status| image:: https://img.shields.io/pypi/v/pdpipe.svg
:target: https://pypi.python.org/pypi/pdpipe

.. |PyPI-Versions| image:: https://img.shields.io/pypi/pyversions/pdpipe.svg
:target: https://pypi.python.org/pypi/pdpipe

.. |Build-Status| image:: https://travis-ci.org/shaypal5/pdpipe.svg?branch=master
:target: https://travis-ci.org/shaypal5/pdpipe

.. |LICENCE| image:: https://img.shields.io/pypi/l/pdpipe.svg
:target: https://pypi.python.org/pypi/pdpipe

.. |Codecov| image:: https://codecov.io/github/shaypal5/pdpipe/coverage.svg?branch=master
:target: https://codecov.io/github/shaypal5/pdpipe?branch=master

Project details


Release history Release notifications

History Node

0.0.25

History Node

0.0.24

History Node

0.0.23

History Node

0.0.22

History Node

0.0.21

History Node

0.0.20

History Node

0.0.19

History Node

0.0.18

History Node

0.0.17

History Node

0.0.16

History Node

0.0.15

History Node

0.0.14

History Node

0.0.13

History Node

0.0.12

History Node

0.0.11

History Node

0.0.10

History Node

0.0.9

History Node

0.0.8

History Node

0.0.7

History Node

0.0.6

History Node

0.0.5

History Node

0.0.4

This version
History Node

0.0.3

History Node

0.0.2

History Node

0.0.1

Download files

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

Filename, size & hash SHA256 hash help File type Python version Upload date
pdpipe-0.0.3.tar.gz (30.5 kB) Copy SHA256 hash SHA256 Source None May 5, 2017

Supported by

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