This directory contains code to reproduce the results from "Inferring influence networks from longitudinal bipartite relational data" (2018) by Frank W. Marrs, Benjamin C. Campbell, Bailey K. Fosdick, Skyler J. Cranmer, and Tobias Böhmelt.
reproduce_blin_methods.R
-- Master file. Run this to reproduce the results from Marrs et. al. (2018).
all_blin_functions.R
-- Load supporting functions and libraries. Requires internet connection to download data and existing code.
countries.txt
-- Countries and their continents used in analysis.
cv_functions.R
-- Cross-validation functions.
MLE_functions.R
-- Functions to estimate BLIN model.
misspec_run_cv.R
-- Cross-validation script.
sid_functions.R
-- Functions to estimate BLIN model for ICEWS data set.
To analyze the ICEWS data, the data must be downloaded. In addition, to estimate the bilinear model, we require the functions from "Multilinear tensor regression for longitudinal relational data" (2015) by Peter D. Hoff. To create the influence network plots, the R package arcdiagram
is required, which is hosted on github (rather than CRAN). Running the master file will automatically download these files (by sourcing all_blin_functions.R
). Lastly, some R packages may be required; all packages are listed in all_blin_functions.R
.