The three main problems addressed by this R package are:

- selecting the most probable structure based on a cache of

pre-computed scores.

- controlling for overfitting.

- sampling the landscape of high scoring structures.

The latter could be very useful in an applied perspective to avoid

reducing the richeness of Bayesian network modelling to report only

**one** structure. Indeed, it allows user to quantify the marginal

impact of relationships of interest by marginalising out over structures

or nuisance dependencies. Structural MCMC seems a very elegant and

natural way to estimate the true marginal impact, so one can determine

if it’s magnitude is big enough to consider as a worthwhile

intervention.

## Description

The package provides a flexible implementation of a structural MCMC sampler for Directed Acyclic Graphs (DAGs). It supports the new edge reversal move from Grzegorczyk and Husmeier (2008) <https://doi.org/10.1007/s10994-008-5057-7> and the Markov blanket resampling from Su and Borsuk (2016) <http://jmlr.org/papers/v17/su16a.html>. It supports three priors: a prior controlling for structure complexity from Koivisto and Sood (2004) <http://dl.acm.org/citation.cfm?id=1005332.1005352>, an uninformative prior and a user defined prior. The three main problems that can be addressed by this R package are selecting the most probable structure based on a cache of pre-computed scores, controlling for overfitting and sampling the landscape of high scoring structures. It allows to quantify the marginal impact of relationships of interest by marginalising out over structures or nuisance dependencies. Structural MCMC seems a very elegant and natural way to estimate the true marginal impact, so one can determine if it's magnitude is big enough to consider as a worthwhile intervention.

mcmcabn is a flexible implementation of a structural MCMC sampler for

Directed Acyclic Graphs (DAGs). It supports the new edge reversal move

from Grzegorczyk and Husmeier (2008)

<https://doi.org/10.1007/s10994-008-5057-7> and the Markov blanket

resampling from Su and Borsuk (2016)

<http://jmlr.org/papers/v17/su16a.html>. It supports three priors: a

prior controlling for structure complexity from Koivisto and Sood (2004)

<http://dl.acm.org/citation.cfm?id=1005332.1005352>, an uninformative

prior and a user defined prior. The three main problems that can be

addressed by this R package are selecting the most probable structure

based on a cache of pre-computed scores, controlling for overfitting and

sampling the landscape of high scoring structures. It allows to quantify

the marginal impact of relationships of interest by marginalising out

over structures or nuisance dependencies. Structural MCMC seems a very

elegant and natural way to estimate the true marginal impact, so one can

determine if it’s magnitude is big enough to consider as a worthwhile

intervention.

-----

## What’s New

- 08/03/2019 - mcmcabn is available on CRAN (v 0.1)

- 18/02/2019 - new pre-print [Is a single unique Bayesian network

enough to accurately represent your

data?](https://arxiv.org/pdf/1902.06641.pdf) on arXiv

- 01/07/2019 - mcmcabn 0.2 available on CRAN

-----

## Future implementations (ordered by urgency)

**`mcmcabn` is developed and maintained by [Gilles

Kratzer](https://gilleskratzer.netlify.com/) and [Prof. Dr. Reinhard

Furrer](https://user.math.uzh.ch/furrer/) from [Applied Statistics

Group](https://www.math.uzh.ch/as/index.php?id=as) from the University

<ahref="#mcmcabn-an-r-package-for-sampling-dags-using-structural-mcmc"class="anchor"></a>mcmcabn: An R Package for sampling DAGs using structural MCMC</h1></div>

<p>(PUBLIC) !!! UNSTABLE VERSION !!!</p>

<p>mcmcabn is a one-man-show (me!) and made of more than 10’000 lines of code which are not bug free! So use it with caution and awareness.</p>

<divid="installation"class="section level2">

<ahref="#mcmcabn-a-structural-mcmc-sampler-for-dags-learned-from-observed-systemic-datasets"class="anchor"></a>mcmcabn: a structural MCMC sampler for DAGs learned from observed systemic datasets</h1></div>

<p>To install <code>mcmabn</code> you need two R packages: <ahref="https://CRAN.R-project.org/package=abn">abn</a> and <ahref="https://CRAN.R-project.org/package=gRbase">gRbase</a> which requires libraries not stored on <ahref="https://cran.r-project.org/">CRAN</a> but on <ahref="http://www.bioconductor.org/">bioconductor</a>. Hence you <strong>must</strong> install these packages <strong>before</strong> installing <code>mcmcabn</code>:</p>

<p>The three main problems addressed by this R package are:</p>

<ul>

<li>selecting the most probable structure based on a cache of pre-computed scores.</li>

<li>controlling for overfitting.</li>

<li>sampling the landscape of high scoring structures.</li>

</ul>

<p>The latter could be very useful in an applied perspective to avoid reducing the richeness of Bayesian network modelling to report only <strong>one</strong> structure. Indeed, it allows user to quantify the marginal impact of relationships of interest by marginalising out over structures or nuisance dependencies. Structural MCMC seems a very elegant and natural way to estimate the true marginal impact, so one can determine if it’s magnitude is big enough to consider as a worthwhile intervention.</p>

<p>The package provides a flexible implementation of a structural MCMC sampler for Directed Acyclic Graphs (DAGs). It supports the new edge reversal move from Grzegorczyk and Husmeier (2008) <ahref="https://doi.org/10.1007/s10994-008-5057-7"class="uri">https://doi.org/10.1007/s10994-008-5057-7</a> and the Markov blanket resampling from Su and Borsuk (2016) <ahref="http://jmlr.org/papers/v17/su16a.html"class="uri">http://jmlr.org/papers/v17/su16a.html</a>. It supports three priors: a prior controlling for structure complexity from Koivisto and Sood (2004) <ahref="http://dl.acm.org/citation.cfm?id=1005332.1005352"class="uri">http://dl.acm.org/citation.cfm?id=1005332.1005352</a>, an uninformative prior and a user defined prior. The three main problems that can be addressed by this R package are selecting the most probable structure based on a cache of pre-computed scores, controlling for overfitting and sampling the landscape of high scoring structures. It allows to quantify the marginal impact of relationships of interest by marginalising out over structures or nuisance dependencies. Structural MCMC seems a very elegant and natural way to estimate the true marginal impact, so one can determine if it’s magnitude is big enough to consider as a worthwhile intervention.</p>

<p>mcmcabn is a flexible implementation of a structural MCMC sampler for Directed Acyclic Graphs (DAGs). It supports the new edge reversal move from Grzegorczyk and Husmeier (2008) <ahref="https://doi.org/10.1007/s10994-008-5057-7"class="uri">https://doi.org/10.1007/s10994-008-5057-7</a> and the Markov blanket resampling from Su and Borsuk (2016) <ahref="http://jmlr.org/papers/v17/su16a.html"class="uri">http://jmlr.org/papers/v17/su16a.html</a>. It supports three priors: a prior controlling for structure complexity from Koivisto and Sood (2004) <ahref="http://dl.acm.org/citation.cfm?id=1005332.1005352"class="uri">http://dl.acm.org/citation.cfm?id=1005332.1005352</a>, an uninformative prior and a user defined prior. The three main problems that can be addressed by this R package are selecting the most probable structure based on a cache of pre-computed scores, controlling for overfitting and sampling the landscape of high scoring structures. It allows to quantify the marginal impact of relationships of interest by marginalising out over structures or nuisance dependencies. Structural MCMC seems a very elegant and natural way to estimate the true marginal impact, so one can determine if it’s magnitude is big enough to consider as a worthwhile intervention.</p>

<li><p>08/03/2019 - mcmcabn is available on CRAN (v 0.1)</p></li>

<li><p>18/02/2019 - new pre-print <ahref="https://arxiv.org/pdf/1902.06641.pdf">Is a single unique Bayesian network enough to accurately represent your data?</a> on arXiv</p></li>

<li><p>01/07/2019 - mcmcabn 0.2 available on CRAN</p></li>

</ul>

<hr>

<p><strong><code>mcmcabn</code> is developed and maintained by <ahref="https://gilleskratzer.netlify.com/">Gilles Kratzer</a> and <ahref="https://user.math.uzh.ch/furrer/">Prof. Dr. Reinhard Furrer</a> from <ahref="https://www.math.uzh.ch/as/index.php?id=as">Applied Statistics Group</a> from the University of Zurich.</strong></p>