This respository contains code to run all experiments for "Inferring dynamic networks from marginals with iterative proportional fitting" by Serina Chang*, Frederic Koehler*, Zhaonan Qu*, Jure Leskovec, and Johan Ugander, published in ICML 2024. The main files are:
- test_ipf.py: functions to run IPF, test and enable IPF convergence (our ConvIPF algorithm), and run the equivalent Poisson regression.
- experiments_with_data.py: functions to run experiments with data - synthetic data, SafeGraph mobility data, and CitiBike bikeshare data
- icml_experiments.ipynb: notebook to recreate all results and figures reported in the paper.
The other files are imported from other repos (Chang et al., 2021) to use SafeGraph mobility data.