Argument crosswalk: psychnet as a substitute for qgraph / IsingFit / mgm / psychonetrics
Source:R/crosswalk.R
net_crosswalk.RdA tidy, one-row-per-argument map from each reference package's estimator to
its psychnet equivalent, so users migrating from qgraph::EBICglasso,
qgraph::cor_auto, qgraph::ggmModSelect, IsingFit::IsingFit,
mgm::mgm, or psychonetrics::panelgvar can find the matching argument and
see what psychnet changes by default. Intensive time-series models
(graphicalVAR, mlVAR) are out of scope; the few-wave panel model is
covered by panel_gvar().
Usage
net_crosswalk(
reference = c("all", "EBICglasso", "cor_auto", "ggmModSelect", "IsingFit", "mgm",
"panelgvar")
)Value
A tidy data.frame, one row per argument, with columns reference
(the pkg::fn being substituted), psychnet (the psychnet verb),
ref_arg, psychnet_arg ("-" when there is no counterpart), status
(identical / renamed / default differs / semantics differ /
reference only / psychnet only), and a short note.
Examples
net_crosswalk("EBICglasso")
#> reference psychnet ref_arg psychnet_arg
#> 1 qgraph::EBICglasso ebic_glasso S cor_matrix
#> 2 qgraph::EBICglasso ebic_glasso n n
#> 3 qgraph::EBICglasso ebic_glasso gamma gamma
#> 4 qgraph::EBICglasso ebic_glasso nlambda nlambda
#> 5 qgraph::EBICglasso ebic_glasso lambda.min.ratio lambda_min_ratio
#> 6 qgraph::EBICglasso ebic_glasso threshold threshold
#> 7 qgraph::EBICglasso ebic_glasso penalize.diagonal penalize_diagonal
#> 8 qgraph::EBICglasso ebic_glasso refit refit
#> 9 qgraph::EBICglasso ebic_glasso checkPD -
#> 10 qgraph::EBICglasso ebic_glasso penalizeMatrix -
#> 11 qgraph::EBICglasso ebic_glasso countDiagonal -
#> 12 qgraph::EBICglasso ebic_glasso returnAllResults -
#> 13 qgraph::EBICglasso ebic_glasso verbose -
#> 14 qgraph::EBICglasso ebic_glasso - data
#> 15 qgraph::EBICglasso ebic_glasso - cor_method
#> 16 qgraph::EBICglasso ebic_glasso - na_method
#> 17 qgraph::EBICglasso ebic_glasso - native
#> 18 qgraph::EBICglasso ebic_glasso - labels
#> 19 qgraph::EBICglasso ebic_glasso - lambda_path
#> status
#> 1 renamed
#> 2 identical
#> 3 identical
#> 4 identical
#> 5 renamed
#> 6 semantics differ
#> 7 renamed
#> 8 semantics differ
#> 9 reference only
#> 10 reference only
#> 11 reference only
#> 12 reference only
#> 13 reference only
#> 14 psychnet only
#> 15 psychnet only
#> 16 psychnet only
#> 17 psychnet only
#> 18 psychnet only
#> 19 psychnet only
#> note
#> 1 correlation matrix in; psychnet also accepts raw data
#> 2
#> 3 EBIC tuning, default 0.5
#> 4 default 100
#> 5 same default 0.01
#> 6 qgraph: logical sig-rule; psychnet: numeric weight cutoff
#> 7 same default FALSE (W_ii = S_ii)
#> 8 qgraph: logical; psychnet also takes "unregularized" (support MLE)
#> 9 psychnet always validates/PD-checks cor_matrix
#> 10 per-edge penalty matrix; psychnet uses scalar rho
#> 11 EBIC df toggle; psychnet never counts the diagonal
#> 12 psychnet always stores $kkt, $lambda, ...
#> 13 psychnet is silent
#> 14 estimate straight from raw data
#> 15 pearson/spearman/kendall/auto
#> 16 pairwise/listwise missing data
#> 17 native solver (TRUE) / glasso Fortran (FALSE)
#> 18 node labels
#> 19 fixed penalty grid, for resampling callers
net_crosswalk("IsingFit")
#> reference psychnet ref_arg psychnet_arg
#> 1 IsingFit::IsingFit ising_fit x data
#> 2 IsingFit::IsingFit ising_fit gamma gamma
#> 3 IsingFit::IsingFit ising_fit AND rule
#> 4 IsingFit::IsingFit ising_fit min_sum min_sum
#> 5 IsingFit::IsingFit ising_fit family -
#> 6 IsingFit::IsingFit ising_fit plot -
#> 7 IsingFit::IsingFit ising_fit progressbar -
#> 8 IsingFit::IsingFit ising_fit lowerbound.lambda -
#> 9 IsingFit::IsingFit ising_fit - nlambda
#> 10 IsingFit::IsingFit ising_fit - lambda_min_ratio
#> 11 IsingFit::IsingFit ising_fit - weights
#> 12 IsingFit::IsingFit ising_fit - na_method
#> 13 IsingFit::IsingFit ising_fit - labels
#> status note
#> 1 renamed binary 0/1 data
#> 2 identical default 0.25
#> 3 renamed AND=TRUE/FALSE -> rule='AND'/'OR'
#> 4 default differs -Inf vs NULL (both keep all rows)
#> 5 reference only psychnet is binomial-only
#> 6 reference only psychnet never plots
#> 7 reference only psychnet has no progress bar
#> 8 reference only psychnet uses its own full lambda path
#> 9 psychnet only lambda-path length
#> 10 psychnet only smallest penalty fraction
#> 11 psychnet only observation weights
#> 12 psychnet only pairwise/listwise
#> 13 psychnet only node labels
net_crosswalk("panelgvar")
#> reference psychnet ref_arg psychnet_arg
#> 1 psychonetrics::panelgvar panel_gvar data data
#> 2 psychonetrics::panelgvar panel_gvar vars vars
#> 3 psychonetrics::panelgvar panel_gvar idvar id
#> 4 psychonetrics::panelgvar panel_gvar beepvar wave
#> 5 psychonetrics::panelgvar panel_gvar missing na_method
#> 6 psychonetrics::panelgvar panel_gvar alpha alpha
#> 7 psychonetrics::panelgvar panel_gvar adjust adjust
#> 8 psychonetrics::panelgvar panel_gvar recursive recursive
#> 9 psychonetrics::panelgvar panel_gvar criterion criterion
#> 10 psychonetrics::panelgvar panel_gvar estimator select
#> 11 psychonetrics::panelgvar panel_gvar nLambda nlambda
#> 12 psychonetrics::panelgvar panel_gvar lambda_min_ratio lambda_min_ratio
#> 13 psychonetrics::panelgvar panel_gvar standardize standardize
#> 14 psychonetrics::panelgvar panel_gvar datatype -
#> 15 psychonetrics::panelgvar panel_gvar within_latent -
#> 16 psychonetrics::panelgvar panel_gvar between_latent -
#> 17 psychonetrics::panelgvar panel_gvar stationary_means -
#> 18 psychonetrics::panelgvar panel_gvar covtype -
#> 19 psychonetrics::panelgvar panel_gvar optimizer -
#> 20 psychonetrics::panelgvar panel_gvar - select
#> 21 psychonetrics::panelgvar panel_gvar - stepup
#> 22 psychonetrics::panelgvar panel_gvar - gamma
#> 23 psychonetrics::panelgvar panel_gvar - refit
#> 24 psychonetrics::panelgvar panel_gvar - labels
#> status
#> 1 identical
#> 2 identical
#> 3 renamed
#> 4 renamed
#> 5 semantics differ
#> 6 identical
#> 7 identical
#> 8 identical
#> 9 identical
#> 10 semantics differ
#> 11 renamed
#> 12 identical
#> 13 semantics differ
#> 14 reference only
#> 15 reference only
#> 16 reference only
#> 17 reference only
#> 18 reference only
#> 19 reference only
#> 20 psychnet only
#> 21 default differs
#> 22 psychnet only
#> 23 psychnet only
#> 24 psychnet only
#> note
#> 1 wide (one row per person) or long data
#> 2 design matrix (variables x waves); psychnet also takes a character vector for long data
#> 3 person id column (long data)
#> 4 wave column (long data); waves ordered by sort()
#> 5 psychonetrics: FIML when data have NA; psychnet: listwise only
#> 6 prune() argument, same default 0.01
#> 7 prune() argument, same default "none"
#> 8 prune() argument, same default FALSE
#> 9 stepup() argument, same default "bic"
#> 10 estimator = "PML" (penalized, auto lambda) is select = "ebic"; FIML is not supported
#> 11 find_penalized_lambda() argument, same default 50
#> 12 find_penalized_lambda() argument, same default 1e-4
#> 13 "none"/"z"/"z_per_wave" as psychonetrics; "quantile" is not offered
#> 14 psychnet infers wide vs long from `vars`
#> 15 psychnet fixes "ggm" (the panelgvar default)
#> 16 psychnet fixes "ggm" (the panelgvar default)
#> 17 psychnet: stationary means (the psychonetrics default)
#> 18 psychnet: ML covariance (denominator n), as psychonetrics uses for raw data
#> 19 psychnet: nlminb + Fisher-scoring polish, ML score stored as a certificate
#> 20 "none" (default) = runmodel() full model with every edge's p-value; "prune" = prune() |> stepup(); "ebic" = PML with EBIC lambda
#> 21 stepup() after prune(); psychnet runs it by default (FALSE = prune() alone)
#> 22 EBIC gamma for select = "ebic" (psychonetrics encodes it in criterion = "ebic.5")
#> 23 unpenalized refit of the EBIC-selected network
#> 24 node labels