Graphitty

A Python library that convert time series to direction Graph to discover the story within data.

Installing

To install this package, runs:

pip install git+https://github.com/sketchytechky/graphitty.git

Remember to add the following line to requirements.txt

git+https://github.com/sketchytechky/graphitty.git

How to use

from graphitty.graphitty import Graphitty
import pandas as pd

# Example input data - see test_simple.py for example
df = pd.DataFrame(
        [...],
        columns=['timestamp','action', 'user_id'])

# Parsing and outputting graph
g = Graphitty(
    df,
    id_col='user_id',
    beahivour_col='action',
    ts_col='timestamp')
nx_graph = g.render()

# create the beautiful directional graph
# see https://github.com/chebee7i/nxpd for usage
import nxpd
nxpd.draw(nx_graph, show='ipynb')

Running Tests

Run test with

py.test --pep8

To run test in watch mode

py.test.watch -- --pep8
# same with: ptw -- --pep8

Related research

• Information Foraging Theory:Adaptive Interaction with Information - http://www.peterpirolli.com/Professional/About_Me_files/IFT%20Ch%201.pdf

• Jakob Nielson example on how to apply information foraging theory to understand visitor's behaviour

  • http://www.useit.com/alertbox/scrolling-attention.html
  • https://www.nngroup.com/articles/information-scent/