Source code by Hongwei Shang, for the paper 'Network A/B Testing: Nonparametric Statistical Significance Test based on Cluster-level Permutation'

For questions, please email Hongwei at hongwei_shang@yahoo.com.

In this directory, we present the Python implementations of computing p-value based on individual-level permutation and cluster-level permutation and generating figures.

Introduction of main files:
- compute.py: compute estimators' p-value at both individual permutation level and cluster permutation level.
- gen_combinations.sh: generate python files with all different combinations, and run python files. 
- result_summary.py: summarize results from simulation study, compute rejection rate at each combination.
- network_ab_plot.ipynb: draw figures in the paper. 


Note: For graph clustering algorithm reLDG, we use the open source code https://github.com/ameloa/streamorder provided by the paper (https://arxiv.org/pdf/2007.03131.pdf)’s author. The streamorder directory is downloaded from https://github.com/ameloa/streamorder with slight modifications. 



