Causal Inference Using Graphical Models with the R Package pcalg

By Markus Kalisch, Martin Mächler, Diego Colombo, Marloes H. Maathuis, Peter Buhlmann. 2012

673 citations recorded by the source corpus.

The pcalg package for R can be used for the following two purposes: Causal structure learning and estimation of causal effects from observational data. In this document, we give a brief overview of the methodology, and demonstrate the package’s functionality in both toy examples and applications.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Causal Generative Neural Networks (2017, methodology): “PC and GES are implemented in the pcalg package [Kalisch et al. , 2012].”
Characterization and greedy learning of interventional Markov equivalence classes of directed acyclic graphs (2011, methodology): “For those algorithms, we enumerated all representative DAGs of the estimated equivalence class using the function allDags() of the R package pcalg (Kalisch et al., 2012), calculated an expression level with each of them, and took the mean of those predictions.”
Order-independent constraint-based causal structure learning (2012, methodology): “Throughout, we assume that all tasks are performed according to the lexicographical ordering of order( V ), which is the standard implementation in pcalg (Kalisch et al., 2012) and TETRAD IV (Spirtes et al., 2000), and is called “PC-1” in Spirtes et al. (2000) (Section 5.4.2.4).”
Causal Structure Learning (2017, methodology): “All methods are called through the interface oﬀered by the CompareCausalNetworks package which depends on the packages backShift (Heinze-Deml, 2017), bnlearn (Scutari, 2010) and pcalg (Kalisch et al., 2012) for the code of the considered methods.”
Bridge Centrality: A Network Approach to Understanding Comorbidity (2019, methodology): “Four networks were DAGs, two which were generated with a Bayesian approach (for details see McNally et al., 2017), and two generated with a PCalg approach (Kalisch et al., 2012).”
Explainable and Interpretable Models in Computer Vision and Machine Learning (2017, neutral): “PC and GES are implemented in the pcalg package Kalisch et al. (2012).”
Partition MCMC for Inference on Acyclic Digraphs (2015, methodology): “While there are a number of software packages, such as pcalg (Kalisch et al., 2012) and bnlearn (Scutari, 2010) in R, for constraint based learning methods, publicly available tools for MCMC based structure learning algorithms are less common, with the possible exception of BDAGL (Eaton and Murphy, ”
MultiVerse: Causal Reasoning using Importance Sampling in Probabilistic Programming (2019, methodology): “Test models and counterfactual queries The 1,000 Structural Causal Models and their corresponding Bayesian networks with binary variables in the form of probabilistic programs were generated similar to the randomDAG procedure in Kalisch et al. (2012). Each Bayesian network contains 15 blocks, where ”
Structural Intervention Distance for Evaluating Causal Graphs (2015, methodology): “It will also be implemented in the pcalg package (Kalisch et al., 2012).”
The Gaussian Graphical Model in Cross-Sectional and Time-Series Data (2016, background): “Several software packages exist that aim to find such a DAG (e.g., pcalg , Kalisch, Mächler, Colombo, Maathuis, & Bühlmann, 2012; bnlearn , Scutari, 2010).”
Qgraph: Network visualizations of relationships in psychometric data (2012, neutral): “…object of class "qgraph" , "loadings" and "factanal" ( stats ; R Development Core Team 2012), "principal" ( psych ; Revelle 2012), "sem" and "semmod" ( sem ; Fox et al. 2012), "lavaan" ( lavaan ; Rosseel 2012), "graphNEL" ( Rgraphviz ; Gentry et al. 2012) or "pcAlgo" ( pcalg ; Kalisch et al. 2012).”
Repair of Partly Misspecified Causal Diagrams. (2017, methodology): “The PC algorithm for de novo structure learning, a two-stage structure learning algorithm available in references 9, 26, is based on tests of conditional independence and is thus agnostic to the causal problem of ultimate interest.”