About

These are scripts written in python to perform statistical analysis on the citation data generated by Webservice. The scripts run multiple tests asserting different hypothesises.

To execute the tests, first setup your development environment:

$ cd [TVQ CLONE PATH]/analytics
$ virtualenv .venv
$ source .venv/bin/activate
# on Windows:
# .\.venv\Scripts\activate
$ pip install -r requirements.txt

All tests are run automatically at the execution of the following command.

python ./lib/run.py ...

You may type the following to see the command help:

python ./lib/run.py --help
usage: run.py [-h] {exe-all,cluster,plot-cluster,t-test,growth-hist,gain-score,pubs-in-clusters,tool-pub,citation-dist} ...

A command-line interface to the scripts implemented for analyzing the TVQ-generated citation data.

positional arguments:
  {exe-all,cluster,plot-cluster,t-test,growth-hist,gain-score,pubs-in-clusters,tool-pub,citation-dist}
                        Commands
    exe-all             Executes all the scripts in a predefined order.
    cluster             Clusters publications in repositories stored in CSV files in given input path.
    plot-cluster        Plot clusters and citation counts in quartiles.
    t-test              Performs t-test on the citation count of publications.
    growth-hist         Plots a histogram of the citation count growth.
    gain-score          Plots gain score.
    pubs-in-clusters    Plots publications in clusters.
    tool-pub            Plots tools in publications.
    citation-dist       Plots citations distribution.

optional arguments:
  -h, --help            show this help message and exit

To see the argument for each sub command you may use the --help argument; for instance:

python ./lib/run.py exe-all --help
usage: run.py exe-all [-h] [-c CLUSTER_COUNT] [-s SOURCE] [-g] [-d] input

positional arguments:
  input                 Path to directory containing input data.

optional arguments:
  -h, --help            show this help message and exit
  -c CLUSTER_COUNT, --cluster_count CLUSTER_COUNT
                        Groups data in the given number of clusters. If not provided, the cluster count is determined automatically using the Elbow method.
  -s SOURCE, --source SOURCE
                        Sets the cluster source.
  -g, --plot_changes    If set, plots changes on clustered citation counts. By default the changes are not plotted.
  -d, --plot_density    If set, plots probability instead of absolute values. Default is False.