Network-based Frequency Analysis (NFA)

The Network-based Frequency Analysis (NFA) tool offers a new methodology to perform frequency analyses using a network approach. In traditional statistics, frequency analyses are performed through histograms where a distribution is split within specific intervals, often called bins, that are not trivial to determine. Network-based frequency analyses take a different approach by looking at ranges around each and every value and create connections when two values are within a certain range of one another. This method has many benefits, including the detection of outliers (since outlying values have no connections). The code can be downloaded by itself below or directly from GitHub. We have also made available a tutorial with a simple numerical example. For citation, please use Derrible, S., Ahmad, N. (2015) Network-Based and Binless Frequency Analyses. PLoS ONE 10(11): e0142108. doi:10.1371/journal.pone.0142108, available here. See our publication page for full details.

version 1.20: GitHub | Download | Tutorial | Publication (for Python 3 and pandas 0.20.3 and later)

version 1.10.1: GitHub | Download | Tutorial | Publication (for Python 2 and pandas 0.18.0 and later – minor bug fix from v1.10)

version 1.10: GitHub | Download | Tutorial | Publication (for Python 2 and pandas 0.18.0 and later)

version 1.00: GitHub | Download | Tutorial | Publication (tested with Python 2 and pandas version 0.14.1)

Network-based Frequency Analysis (NFA)