mdaquin / benford_wikidata Goto Github PK

This includes scripts to test how well properties with numerical values on wikidata fit benford's law. To reconstruct the results, download the latest.all.nt.gz file from wikidata dumps, run extract.sh, and concatenate the results into a file called "numbers.nt".

Then:

• nseries.py create a json files with the relevant stats for each properties
• label.py find the labels and descriptions of each property
• fit.py run the chi squared test for each property against Benford's law
• stats.py creates a csv file with all the results.