v2.4.3

v2.4.3 (18 November 2024)

• New module to convert igraph graph objects to/from ALAAMEE internal format
• Added insertEdge() as alias for insertArc() on Digraph so that it can be used consistently for Graph, BipartiteGraph and Digraph objects
• run_sa() and run_ee() functions in estimateALAAMSA and estimateALAAMEE modules respectively to run estimation with supplied Graph (or Digraph or BipartiteGraph) object (which contains attributes, rather than supplying attributes as filenames of attribute data files)

Full Changelog: v2.4.2...v2.4.3

v2.4.2

v2.4.2 (12 August 2024)

• Print summaries of NA values by mode in bipartite networks
• Added outcome with NA set to 0 for persons and scale and centre centrality measures in Evtushenko director interlock data example
• Allow Ainitial to be specified in simulation for bipartite networks
• Fixed bug with not passing bipartiteGoFfixedMode in estimateALAAMSA through to simulation function
• Added Seierstad & Opsahl (2011) Norwegian company boards example
• Added St Louis crime data (Decker et al. 1991) example
• Fixed bug with getting random node in mode B in bipartite graphs
• Implemented outputGoFstatsFilename and outputObsStatsFilename parameters in estimateALAAMSA.run_on_network_attr() for gof()
• Implemented new parameters bipartiteFixedMode, outputStatsFilename, outputObsSttatsFilename and labels in gof()
• Scripts to convert SA output to LaTeX now handle Mahalanobis distance from GoF output
• Implemented new change statistics for two-paths with matching or mismatching categorical attributes: changeSamePartnerActivityTwoPath, changeDiffPartnerActivityTwoPath, changeSameIndirectPartnerAttribute, changeDiffIndirectPartnerAttribute, changeAlterBinaryTwoStar1, changeAlterBinaryTwoStar2, changeBpAlterSameTwoStar1, changeBpAlterDiffTwoStar1, changeBpAlterSameTwoStar2, changeBpAlterDiffTwoStar2, changeBpAlterBinaryTwoStar1, changeBpAlterBinaryTwoStar2
• Increased parameter name field with in formatted output in estimateALAAMSA.run_on_network_attr()
• Implemented new --shading (-s) option on plotSimulationDiagnostics.R to plot shaded area on trace plots for 95% interval, which also means using PDF instead of EPS when this option is used
• New and modified scripts for more simulation experiments for revised manuscript, use max. 500 runs
• This is the version used for the journal revision of manuscript arXiv:2404.03116

Full Changelog: v2.4.1...v2.4.2

v2.4.1

v2.4.1 (25 March 2024)

• Added GEMSEC Deezer examples (as used in new manuscript Stivala, Wang, & Lomi (2024) "ALAAMEE: Open-source software for fitting autologistic actor attribute models")

• Added scripts to convert multiple EE estimations to LaTeX table and single SA estimation to LaTeX table

Full Changelog: v2.4.0...v2.4.1

v2.4.0

v2.4.0 (8 January 2024)

• Implemented BipartiteGWActivity change statistic
• Implemented bipariteFixedMode on simulation
• Added -e (--eiindex) option for Krackhardt & Stern (1988) E-I index on plotALAAMEEsimfit.R
• Added ALAAM examples with badger data from Silk et al. (2017)
• Implemented oO_OsameContagion change statistic
• Implemented distance matrix and ContagionDist and GeographicHomophily change statistics
• Added Hamilton et al. (2020) Tasmanian Devil replication data and TERGM models, and new models for the Tasmanian Devil data using SAOM (RSiena) and ALAAM (ALAAMEE stochastic approximation)
• Added test cases and regression tests for new change statistics
• Added experimental change statistics LogContagion and PowerContagion including high school example models and simulation experiments scripts
• Added Mahalanobis distance to goodness-of-fit, including test cases for Mahalanobis distance computation
• Added add_gof_param_func_list parameter to estimateALAAMSA() to allow user-specified goodness-of-fit statistics

Full Changelog: v.2.3.4...v2.4.0

v2.3.4

This release includes the new geometrically weighted effects GWActivity, GWSender, and GWReceiver (see manuscript cited below), as well as EgoInThreeStar and EgoOutThreeStar and python functions to parse the output from computeALAAMEECovariance.R and use it to automate the simulation-based goodness-of-fit process.

Also includes scripts to get and convert data and run simulation experiments for the manuscript cited below (and more), as well as adding degree distribution nodes with outcome attribute and attribute assortativity to goodness-of-fit plots, and option to output the simulated vectors not just the statistics from simulations.

Changed t-ratio threshold in stochastic approximation gof output format script to 2.0.

This release was used for results in manuscript Stivala, A. (2023). Overcoming near-degeneracy in the autologistic actor attribute model. arXiv preprint arXiv:2309.07338. https://arxiv.org/abs/2309.07338

Full Changelog: v2.3.3...v.2.3.4

v2.3.3

This release adds various examples and scripts used for manuscript about ALAAMEE.
Added the -r (--max-runs) option to computeALAAMEEcovariance.R, used for generating some results used in manuscript.
Replace some change stats with faster (but more verbose and less elegant) versions; others retained and used in regression tests.

Full Changelog: v2.3.2...v2.3.3

v2.3.2

Implemented more change statistics for directed networks: Reciprocity, EgoIn2Star, EgoOut2Star, Mixed2Star, Mixed2StarSource, Mixed2StarSink, ContagionReciprocity, TransitiveTriangleT1, TransitiveTriangleT3, TransitiveTriangleD1, TransitiveTriangleU1, TransitiveTriangleC1, TransitiveTriangleC3, AlterInTwoStar2, AlterOutTwoStar2, SenderMatch, ReceiverMatch, ReciprocityMatch, SenderMismatch, ReciprocityMismatch.

Includes documentation in comments for each statistic (including ASCII diagram) and test cases verified against MPNet.

Full Changelog: v2.3.1...v2.3.2

v2.3.1

Uses sparse matrix for two-paths lookup matrix for more efficient and scalable change statistic computation.
Add bipariteGoFfixedMode to stochastic approximation estimation so that for bipartite networks, can fix either mode A or B nodes if the outcome variable is not defined for that mode.
Improvements to output formatting for stochastic approximation estimations.
More regression tests, including for bipartite change statistics.
Includes R scripts to load, merge, convert, and get centrality scores for director interlock data.
Fix bug in changeo_Ob and contagion with NA_VALUE.
Rename files/functions from simpleDemo to estimateALAAM etc. and update comments etc.
Added changeBipartiteEgoThreeStar change statistic.
Added bipartite Goodness-of-Fit example.

Full Changelog: v2.3...v2.3.1

v2.3

Version 2.3 includes support for two-mode (bipartite) graphs.
Now requires Python 3.9, as it uses the @functools.cache decorator to memoize a function. Note if you remove this decorator, it will still work with version 3.8 (required to use the := assignment operator in an expression).

v2.2

Version 2.2 includes implementation of ALAAM parameter estimation (stochastic approximation or equilibrium expectation), simulation, and goodness-of-fit for directed networks. (Only a limited number of change statistics for directed networks implemented so far).

Full Changelog: v2.1...v2.2