PCABM

This is the sample code for conducting pairwise covariates-adjusted stochastic block model (PCABM), which is a generalization of SBM that incorporates pairwise covariate information.

Simulation Example.ipynb is a a toy example illustrating how to use pcabm package, with data generating from a pcabm. Political Blog Example.ipynb is applying mle of pcabm to a famous real world data set.

Followings are introductions to pcabm's files:

• commFunc.py : some common functions that will be used often.
• dcbm.py : degree corrected block model
• pcabm.py : pairwise covariates-adjusted stochastic block model using tabu search
• sc.py : different spectral clustering methods
• ecv.py : using edge cross validation to choose K and covariates
• plem.py : pairwise covariates-adjusted stochastic block model using pseudo likelihood

To replicate simulation results, people could run script in 'simulations' folder. The name is the figure number in the paper. To run a single simulation, use

python filename.py

To run multiple simulations, use

sbatch filename.sh

To save the simulation results, you need to make a folder './output/filename', or anywhere else you'd like. To aggregate simulation results, use corresponding commented command in filename.sh.