jhart96 / bisonr

An R package for Bayesian Inference of Social Networks

Hi there, I'm Jordan 👋

I'm a researcher and data scientist interested in Bayesian modelling and machine learning.

bisonr's Issues

Add R functions for setting custom priors

Function to extract global network metrics

Improve simulated datasets

Generate simulated data with multi-membership effects for dyadic regression and random effects for edge models, dyadic regression, and nodal regression.

ICL comparison mixture models

Default priors aren't set for conjugate models.

Get Github Actions to automatically run `install_cmdstan`

Support for multiple networks in nodal and dyadic regression

Move Stan models to separate files again

Network metrics

• Clustering coefficient

Partial pooling model

Get successful R CMD check

Zero-inflated edge models

Fix array assignment notation

Write documentation

Prior check function

bison_to_igraph node ordering

Align igraph vertices with bison node_to_idx

Prior predictive checks

Add support for directed networks

Function to return mixture component locations.

Add random effect support for dyadic and nodal regression models

Option to conditionally filter rows in bison_brms

Cox survival model

Use improved formula parser for dyadic and nodal regression

bison_brm get prior wrapper

Add model comparison function

Automatically assign factors to node IDs

Write examples for main use cases

Add duration model

Write unit tests

Convert from GBI function

Add support for missing dyads

Add random effect support for duration model

Mixture model support

Add conjugate posterior method for binary and count methods.

Use index instead of design matrix for edge weight parameters

Improve Documentation

Add support for user-specified priors

R-hat brms warning

Support edge weights/metrics as predictors

Transform argument for posterior draws

Check approximation accuracy for multivariate normal.

Add filter option to plot_network

Correlated effects directed network

Fix division by zero issue in conjugate priors

Change summary() behaviour

Combine fixed and mixed effect models into a single model with conditionals

Function to plot mixture model components.

Update and combine Stan models

• #3
• #4
• #5
• #7
• #8
• #17

Change simulation column names from event/duration.

Discrete metrics should be plotted with histogram