easybgm in jasp #124
easybgm in jasp #124
Conversation
… of the prior options properly.
…e centrality measures when model = ggm || gcgm; added a note that the omrf is slow.
|
hurray, well done! |
| export(BayesianNetworkAnalysis) | ||
| export(NetworkAnalysis) |
There was a problem hiding this comment.
did roxygen2 do this automatically? if yes all good, otherwise please revert the change.
There was a problem hiding this comment.
It was changed automatically.
R/bayesiannetworkanalysis.R
Outdated
| easybgmFit <- try(easybgm::easybgm(data = as.data.frame(apply(dataset[[nw]], 2, as.numeric)), | ||
| type = "continuous", | ||
| package = "BDgraph" , | ||
| iter = as.numeric(options[["iter"]]), | ||
| save = TRUE, | ||
| centrality = FALSE, | ||
| burnin = as.numeric(options[["burnin"]]), | ||
| g.start = options[["initialConfiguration"]], | ||
| df.prior = as.numeric(options[["dfprior"]]), | ||
| g.prior = as.numeric(options[["gprior"]]))) |
There was a problem hiding this comment.
| easybgmFit <- try(easybgm::easybgm(data = as.data.frame(apply(dataset[[nw]], 2, as.numeric)), | |
| type = "continuous", | |
| package = "BDgraph" , | |
| iter = as.numeric(options[["iter"]]), | |
| save = TRUE, | |
| centrality = FALSE, | |
| burnin = as.numeric(options[["burnin"]]), | |
| g.start = options[["initialConfiguration"]], | |
| df.prior = as.numeric(options[["dfprior"]]), | |
| g.prior = as.numeric(options[["gprior"]]))) | |
| easybgmFit <- try(easybgm::easybgm(data = dataset[[nw]], | |
| type = "continuous", | |
| package = "BDgraph" , | |
| iter = options[["iter"]], | |
| save = TRUE, | |
| centrality = FALSE, | |
| burnin = options[["burnin"]], | |
| g.start = options[["initialConfiguration"]], | |
| df.prior = options[["dfprior"]], | |
| g.prior = options[["gprior"]])) |
so in principle, the data should already be in the right format at this point (e.g., a data.frame with only numeric values). In addition, the values from the options should also be actual values already. So the removed lines should be redundant. If these lines are somehow not redundant, then this should probably be fixed when reading the data.
There was a problem hiding this comment.
I fixed most of the arguments. Except for transforming the variables to numeric when model == "ggm" || "gcgm". This is because if you pass non-numeric variables to easybgm when using BDgraph, an error occurs internally in the get_S_n_p function. I did not want to alter the .networkAnalysisReadData function, which is used for all network analyses. However, I am open to suggestions if you think this needs to be handled differently.
sekulovskin
force-pushed
the
easybgminJASP
branch
from
cb036cb to
2ea84cb
Compare
…orkAnalysisNetworkPlot function to handle estimator/model based on the selected method; (2) corrected the silly loop when estimating the omrf; (3) deleted somestimating the omrf; (3) deleted some redundant comments; (4) changed the footnote under the table. Round 2 with the rest of the changes will follow soon.
|
Should be this in 0.19.2, or can it wait until the release afterwards? |
…, adjusted the arguments in functions, and some other small fixes.
Merge branch 'master' into easybgminJASP # Conflicts: # R/bayesiannetworkanalysis.R # inst/qml/BayesianNetworkAnalysis.qml
Implemented the easybgm package in JASP, which now works in place of BDgraph and can handle: ggms and gcgms (using BDgraph in the background) as well as the omrf (using bgms in the background).
There are also some other cosmetic changes like added footnotes and prior options.
For now, I only allow the software to compute credible intervals for the centrality measures when model == omrf.
The B-C model is not included yet, I will probably do that next week.