Skip to content

Latest commit

 

History

History
198 lines (125 loc) · 7.58 KB

NEWS.md

File metadata and controls

198 lines (125 loc) · 7.58 KB
Raw

sperrorest 3.0.5

  • adjust handling of class name in response to R check 'Note'

sperrorest 3.0.4.9000

  • adjust handling of class name in response to R check 'Note'

sperrorest 3.0.4

  • implemented parallelization modes at fold and repetition levels
  • more flexible permutation options (especially for LOO CV)
  • handle formulas with 0 or 1 predictor(s), e.g. y~1
  • updated vignette

sperrorest 3.0.3.0001

  • "fixed train_fun / test_fun issue
  • more flexible parallelization of add.distance

sperrorest 3.0.3

  • added parallelization modes (arguments mode_rep and mode_fold)
  • more flexible permutation (argument imp_sample_from)
  • updated vignette

sperrorest 3.0.2.9000

  • partition_disc(): set default value of arg buffer to 0 instead of NULL, fixes #61

sperrorest 3.0.2

  • partition_disc(): set default value of arg buffer to 0 instead of NULL, fixes #61
  • partition_loo(): Sequence along observations instead of columns. Before, the train set was only composed of ncol observation. (#60)

sperrorest 3.0.1

  • Resolve minor NAMESPACE issues for {future} on CRAN

sperrorest 3.0.0

Breaking

  • All parallel modes got removed in favor of using parallelism via package {future}. This simplifies usage a lot and helps simplifying future maintenance. Also it gives users the freedom to choose the parallel backend on their own.
  • sperrorest() run sequentially by default again rather than in parallel.

Package infrastructure

  • Removed the following packages from Imports: {glue}, {purrr}, {doFuture}, {gdata}, {magrittr}, {parallel}, {pbapply}, {pbmcapply}, {foreach}
  • Moved the following packages from Imports to Suggests: {rpart}
  • Removed the following packages from Suggests: {pacman}, {ipred}, {RSAGA}
  • The "spatial-use-case" vignette now runs and is not read-only anymore.

sperrorest 2.1.5 (20-Mar-2018)

Bugfixes

  • Updated error message displayed to the user if err_fun() throws an error during performance calculation. An exemplary case would be a binary classification in which only one level of the response exists in the test data (due to spatial partitioning).

sperrorest 2.1.4 (12-Feb-2018)

Minor

  • import future_lapply from future.apply instead of future

sperrorest 2.1.3 (08-Dec-2017)

Minor

  • update CITATION file

sperrorest 2.1.1 (15-Oct-2017)

Bugfixes

  • train_fun and test_fun are now handled correctly and eventual sub-sampling is correctly reflected to the resulting 'resampling' object

sperrorest 2.1.0 (25-Sep-2017)

Features

  • error handling during model fitting & performance evaluation: If a model does not converge for some folds or an error occurs during performance calculation, results of this fold are set to NA and a message is printed to the console. sperrorest() will continue normally and uses the successful folds to calculate the repetition error. This helps to run CV with many repetitions using models which do not always converge like maxnet(), gamm() or svm().

Bugfixes

  • Size of example data set ecuador has been adjusted to avoid exact duplicates of partitions when using partition_kmeans().

sperrorest 2.0.1 (20-Jul-2017)

Bugfixes

  • Fixes a bug which caused equal importance of all predictors when performing permutation-based variable importance assessment

sperrorest 2.0.0 (12-Jun-2017)

Major

  • integration of parsperrorest() into sperrorest().
  • by default, sperrorest() now runs in parallel using all available cores.
  • runfolds() and runreps() are now doing the heavy lifting in the background. All modes are now running on the same code base. Before, all parallel modes were running on different code implementations.
  • function and argument name changes to 'snake_case'

Features

  • new (parallel) modes:
    • apply: calls pbmclapply() on Unix and pbapply() on Windows.
    • future: calls future_lapply() with various future options (multiprocess, multicore, etc.).
    • foreach: foreach() with various future options (multiprocess, multicore, etc.). Default option to cluster. This is also the overall default mode for sperrorest().
    • sequential: sequential execution using future backend.
  • RMSE instead of MSE as error measure
  • You can now pass also single values to repetition argument of sperrorest(). Specifying a range like repetition = 1:10 will also stay valid.
  • New vignette sperrorest::parallel-modes comparing the various parallel modes.
  • New vignette sperrorest::custom-pred-and-model-functions explaining why and how custom defined model and predict functions are needed for some model setups.

Misc

  • Limit workers to number of repetitions if number of cores > number of repetitions. This ensures that no unnecessary workers are started and increases robustness of parallel execuction.
  • documentation improvements.
  • do_try argument has been removed.
  • error.fold, error.rep and err.train arguments have been removed because they are all calculated by default now.

Bugfixes

  • partial matching of arguments
  • account for factor levels only present in test data but missing in training data. Previously, {sperrorest} errored during the predict step when this occurred. Now, this is accounted for and an informative message is given.

sperrorest 1.0.0 (08-Mar-2017)

New features

  • add parsperrorest(): This function lets you execute sperrorest() in parallel. It includes two modes (par.mode = 1 and par.mode = 2) which use different parallelization approaches in the background. See ?parsperrorest() for more details.

  • add partition.factor.cv(): This resampling method enables partitioning based on a given factor variable. This can be used, for example, to resample agricultural data, that is grouped by fields, at the agricultural field level in order to preserve spatial autocorrelation within fields.

  • sperrorest() and parsperrorest(): Add benchmark item to returned object giving information about execution time, used cores and other system details.

Changes to functions:

  • {sperrorest}(): Change argument naming. err.unpooled is now error.fold and err.pooled is now error.rep

  • sperrorest() and parsperrorest(): Change order and naming of returned object

    • class sperrorestpoolederror is now sperrorestreperror
    • returned {sperrorest} list is now ordered as follows:
      1. error.rep
      2. error.fold
      3. importance
      4. benchmarks
      5. package.version

Package related

sperrorest 0.2-1 (19 June 2012)

  • First release on CRAN

sperrorest 0.2-0 (19 June 2012)

  • last pre-release version
  • replaced Stoyan's data set with Jannes Muenchow's data, adapted examples

sperrorest 0.1-5 (1 Mar 2012)

  • made training set estimation optional
  • robustified code using try()

sperrorest 0.1-2 (29 Jan 2012)

  • internal release 0.1-2
  • some bug fixes, e.g. in err.* functions
  • improved support of pooled versus unpooled error estimation
  • changed some argument names
  • this version was used for Angie's analyses

sperrorest 0.1-1 (29 Dec 2011)

  • built internal release 0.1-1

sperrorest 0.1

  • general code development (2009 - 2011)
  • package project and documentation created (Oct-Dec 2011)