The package has undergone a major make-over. A slight, but breakable, change in the api of fit_outlier. The documentation of fit_outlier has been updated and now includes more and better examples of how and when to use the function. The fit_graph function is no longer a part of molic. It now lives in its own package at ess and molic is now dependend on ess. It is therefore now required to run include(ess) to have access to fit_graph.
The readme file has also undergone a major change - the former example using cars data has been removed; it was never really a good example showing how to do outlier detection with molic.
derma has been included and a new vignette using this data has been added.tgp_dat data has now been compressed to save disk space.plot.gengraph function applied to an object (gengraph) returned from one of the graph fitting functions (fit_graph, fit_components etc.) now takes an input that let the user specify the color of the nodes.subgraph function is now provided.sapply's are now converted to vapply's for safety and potentially more speed when fitting graphs.pmf no longer plots the density of the deviances of a outlier_model object. Use plot for this instead; this is now consistent with the other related functions like fit_outlier. Instead pmf is used to construct the probability mass function of a decomposable graphical model which can be used to obtain probabilities of observing specific cells/observations/configurations.Development Model
From this release we adopt the branching model introduced by Vincent Driessen
This means, that there are now two branches: the master branch is always the current stable version, and the develop branch is the develop version.
New API
fit_outlier that depends on an adjacency list no accept gengraph objects returned from fit_graph - i.e. no need to use adj_lst() first.New functions
generate_multiple_models
fit_graph and fit_outlier that conducts all the hypothesis \(H_k:\) \(y\) has level \(k\) for \(k = 1,2,\ldots, l\).plot.multiple_models
fit_multiple_models this function is used to visualize all the hypothesis tests for a single observation simultaneously. It is a ggplot2 objectplot.outlier
fit_outlier this function is used to visualize the approximated density of the deviance under the null hypothesis. It is a ggplot2 object.components
fit_componentsMisc * All deviances are now non-negative as they should be! Before, a constant was neglected which could potentially confuse the users since a deviance is per definition non-negative.