bonvoyage

What is bonvoyage?

Transform percentage-based units into a 2d space to evaluate changes in distribution with both magnitude and direction.

• Free software: BSD license
• Documentation: https://yeolab.github.io/bonvoyage

Installation

To install anchor, we recommend using the Anaconda Python Distribution and creating an environment, so the anchor code and dependencies don’t interfere with anything else. Here is the command to create an environment:

conda create -n anchor-env pandas numpy matplotlib seaborn scikit-learn

Stable (recommended)

To install this code from the Python Package Index, you can install on the command line via pip:

pip install bonvoyage

Bleeding-edge (for the brave)

To install this code, clone this github repository and use pip to install

git clone git@github.com:yeolab/bonvoyage
cd bonvoyage
pip install .  # The "." means "install *this*, the folder where I am now"

Usage

To use bonvoyage to get waypoints, you want your data to be a pandas DataFrame of shape (n_samples, n_features)

import bonvoyage

wp = bonvoyage.Waypoints()
waypoints = wp.fit_transform(data)

bonvoyage is modeled after scikit-learn in is method of creating a transforming object and then running fit_transform() to perform the computation.

To plot the waypoints, use a waypointplot, which can do either "scatter" or "hex" plot types. By default, hexbin plots are used:

import bonvoyage

bonvoyage.waypointplot(waypoints)

You can also specify to use scatter:

import bonvoyage

bonvoyage.waypointplot(waypoints, kind='scatter')

To add color, give a series or other groupby-able object:

import bonvoyage

bonvoyage.waypointplot(waypoints, kind='scatter', features_groupby=modalities)

History

1.0.0 (2017-06-28)

• Added tests and examples

0.1.0 (2015-09-15)

• First release on PyPI.