Statistical Experiments on Batch Computing Clusters
Extends the BatchJobs package to run statistical experiments on
batch computing clusters. For further details see the project web page.
- Added support for read-only mode of registries. See ?loadRegistry for more
- Compability with new versions of testthat, RSQLite and DBI.
- Problem and algorithm ids may now contain the dot char (".").
- Fixed a bug in summarizeExperiments where column "repl" was not correctly
handled in the output
- Added convenience function getExperimentParts()
- findExperiments now optionally supports regular expression matching
- reduceResultsExperiments can now apply a function on jobs with
missing results if explicitly requested
- fixed a minor bug in summarizeExperiments where a column was not named as
- updated citation information
- Updated to use RSQLite >=1.0.0
- Argument 'progressbar' added to multiple functions to suppress displaying the
- Fixed a bad bug in reduceResultsExperiments: This bug was only triggered when argument
impute.val was used. It could have lead to wrong 'id' column in result data.frame, no
other data in the result as corrupted.
- Support impute.val in reduceResultsExperimentsParallel.
- Reduced disk i/o in chunks
- New argument 'ids' (job ids) for getProblemIds and getAlgorithmIds
- New argument 'impute.val' for reduceResultsExperiments
- Support for sourcing of files and directories
- fixed a bug regarding addExperiments and reordering of parameters
- print method for designs now contains parameter storage informations
- new helper: getJobInf, reduceResultsExperimentsParallel
- fix for case-insensitive file systems
- documentation fixes
- various smaller fixes
- added missing import for R versions < 2.15.1
- reworked summarizeExperiments, the previous "details" options was basically unuseable
- minor bug fixes