This Python package is intended to allow Zooniverse Project Owners to quickly run some simple analysis on the classification CSVs that the Zooniverse backend allows you to export via the Data Exports page.
Download or clone the GitHub repo, then install it with pip.
$ cd pyniverse/
$ python -m pip install .
For development work, install the linting, formatting, and test tools too:
$ python -m pip install -e ".[dev]"
Most of the logic in Pyniverse is hidden away in a simple class, called Classifications, which contains a variety of methods, including several that plot graphs. The installed CLI command creates an instance of the class by passing it the path of the CSV file downloaded from Zooniverse and calling several of the methods. Let's see how it works.
$ cd examples/
$ zooniverse-classifications-analyse --input_file dat/test-zooniverse-classifications.csv.bz2
Reading classifications from CSV file...
Total classifications: 218629
Total users: 4529
Gini coefficient: -0.78
Top 10 users have done: 18.6 %
Top 100 users have done: 44.4 %
Top 1000 users have done: 82.8 %
This step should take no more than 30 seconds and in addition to the above information, you should find some graphs in graphs/. If you didn't specify the name of the output file using the --output_stem option then the program will use the default which is test.
$ ls graphs/
test-classifications-day.pdf test-classifications-week.pdf test-user-distribution.pdf test-users-week.pdf
test-classifications-month.pdf test-users-day.pdf test-users-month.pdf
There are three main graphs produced. The first is simply the number of classifications against time. Three time periods are produced: by day, by week and by month and a cumulative line is added.
The next is the number of users trying the project for the first time, again by day, by week and by month.
And lastly the cumulative user distribution so you can see how asymmetric the contribution of the users is.
If you use this package, please cite it using the DOI below