Release of MplusAutomation 1.2

[michael.hallquist]

Dear MplusAutomation community,

I wanted to share the MplusAutomation 1.2 is now available on CRAN. In addition to a number of bugfixes and small improvements, this is a major release that has three substantial updates.

1) I've introduced submitModels, used for submitting large batches of models on high-performance computing clusters. This is akin to runModels, but where you have many models you'd like to estimate asynchronously. For more details, see: https://michaelhallquist.github.io/MplusAutomation/articles/submitModels.html.

2) I've introduced mplusModel (not to be confused with mplusModeler), an R6 class that streamlines inline integration of Mplus and R. For example if you provide an inp file to it, like mplusModel(inp='myfile.inp'), it will look for and load the corresponding output using readModels and also read the dataset into the $data element. The structure of the object is very similar to what is given by readModels, but you now have access to the methods $run and $submit, which call runModels and submitModels, respectively. This is still work in progress, but in the long run, mplusModel will replace mplusModeler and mplusObject.

3) I've improved the guts of readModels such that the parsing time is about 66% less. In addition, if you have large SAVEDATA files (e.g., FSCORES) or BPARAMETERS, the parser for those is now 10x faster.

The full list of updates is below.

Cheers,

Michael

--

Version 1.2
 - notice:      MplusAutomation examples vignette updated to refer to readModels, not outmoded extract* functions
 - feature:     New pathfilter argument to readModels allows for selection of outputs to read based both on path name (pathfilter) and file name (filefilter)
 - feature:     compareModels now returns a list of comparison results including chi-square difference tests when requested
 - feature:     New mplusModel object that simplifies inline integration of Mplus and R.
 - feature:     New submitModels function supports asynchronous submission of Mplus models on high-performance clusters using slurm or torque
 - feature:     Rewrote parser for fixed-width Mplus output objects (esp. SAVEDATA files). Now 10x faster!
 - feature:     Use data.table fread for reading delimited files (e.g., BPARAMETERS). Now 10x faster!
 - feature:     Rewrote parameter chunk parser and adjoining functions for speed, reducing overall time for readModels by approximately 66%!
 - feature:     Parser for SAVEDATA outputs now handles massive records where Mplus splits records across rows (esp. SAVE = FSCORES)
 - feature:     readModels extracts odds ratios and their confidence intervals into $parameters$odds and $parameters$ci.odds, respectively. Addresses #192.
 - feature:     readModels extracts results of the R3STEP procedure in $r3step. Closes #204
 - feature:     readModels extracts new H5RESULTS object from Mplus v8.11 in $h5results
 - feature:     compareModels now returns a list containing the results of the comparison, including the diffTest. Closes #185.
 - notice:      Package namespace has been cleaned up to remove long-deprecated functions including extractModelSummaries, extractModelParameters, and extractModIndices.
 - bugfix:      readModels now detects filefilter arguments that use $ or ^ to denote the end or beginning of a string, respectively, and also handles ".out" being included in filefilter Addresses #194.
 - bugfix:      readModels now parses indirect effects for NB/ZI models with binary outcomes annotated with #. Closes #217.