Autologistic regression modelling. MATLAB code supporting the paper "Better Autologistic Regression"
This repository contains the MATLAB code and data files written in support of the paper:
Mark A. Wolters (2017), Better Autologistic Regression, Frontiers in Applied Mathematics and Statistics, 3(24), 1-20, https://doi.org/10.3389/fams.2017.00024.
Please cite this paper if you make use of this repository.
Autologistic regression is a statistical model for analysis of correlated binary responses. For details, see the above-linked article. An online demo can also provide some background. Using the code in this repository, one can fit autologistic regression models and also explore the characteristics of such models, e.g. by drawing samples from them.
While considerable effort has been made to ensure that the software is correct and useful, it is research-level software. A more comprehensive and performant implementation is currently under development (using the Julia language) at the time of writing.
Brief Description of the Code
The repository consists of two folders.
ALRclasses contains MATLAB OOP code implementing the autologistic models. It should be on your search path.
Create a model by instantiating an object and setting its members (parameters, covariates, responses, etc.). Parameter estimation can be done using matlab optimization functions.
PaperCode contains scripts, functions, and data files used to do the analyses in the paper. The scripts have filenames beginning with
Script_. A good place to start is
Copyright 2018 Mark Anthony Wolters. This repository is under the MIT license. See the license file.