CRAN release: 2023-09-13
- Fix: Parquet datasets now work where columns simultaneously have to be renamed and converted.
- Added upload functions:
publish_datato upload a dataset on OpenML
publish_taskto create a task on OpenML
publish_collectionto create a collection on OpenML
CRAN release: 2023-07-06
- Listing functions don’t return the tables invisibly anymore.
- Address CRAN NOTE regarding unused bit64 import.
- Improved the printer for all OpenML objects.
benchmark_grid_oml(), which was already deprecated in release 0.7.2.
- Removed the fields
OMLCollectionclass. Consequently, the
cacheoption can no longer be set for
OMLCollectionobjects, see the class documentation for more information.
- Removed the examples, as they caused problems with CRAN checks when OpenML was unavailable.
- Caching can no longer be specified at the instance level but only globally through the option
$download()method for all OML objects to fully download an object for offline usage.
- Incremented the cache version for parquet data due to a change in OpenML.
- Added an online tutorial for the package.
CRAN release: 2023-06-12
- Fix: target is added to features when converting a
OMLDataobject to a task with an explicit target variable that is not the default target.
benchmark_grid_oml()in favour of
mlr3::benchmark_grid(..., paired = TRUE)
- Fix: Incremented cache version for data objects for int64 data types (introduced in the previous release).
- Fix: Incremented cache version for data description and fixed bug, as
make.names()was not applied to ignore attributes.
- Fix bug in task converter (features were sometimes not set correctly)
- Collection now shows name in printer
- Better error message when parquet dataset creation fails
CRAN release: 2023-03-14
- Fixed argument names of S3 method for
as_data_backendto comply with new CRAN checks
CRAN release: 2022-12-12
- feature: Add argument
- fix: strings and nominals are distinguished for parquet files
- docs: Fixed some OpenML links
- docs: Renamed the docs for OpenML objects
- Renamed the sugar functions from:
- Addresses a CRAN issue: examples fail gracefully if OpenML server is busy.
CRAN release: 2022-10-12
- Add R6 classes for
- Added function
benchmark_grid_omlthat allows for easier creation of benchmark designs from OpenML task-resampling pairs.
- Added sugar functions
oml_collectionfor all OpenML objects.
- Conversion from OpenML to mlr3 objects is now only possible with the usual s3-converters
as_<object>. This improves consistency by ensuring that the subcomponents of OpenML objects are always OpenML objects and not suddenly mlr3 objects.
- Added more converter functions:
- Added support for parquet files that were recently introduced on OpenML. The global option
mlr3oml.parquetcan be used to enable or disable this. By default it is
FALSE. This is implemented via the duckdb backend from
- Support to use the OpenML test server. This can be globally enabled using the option
mlr3oml.test_serveror individually for objects. Options to globally define an API-key for the test server are through the environment variable
TESTOPENMLAPIKEYor the option
- Removed support for survival tasks as mlr3proba is no longer on CRAN
- OpenML tasks can now also be filtered according to the task type
- Implement an arff writer and remove the arff dependency, therefore also removing the option
- Increment the cache version number due to changes in the cache structure: This will flush the previous cache folder.
- Simplified the code structure by adding
OMLObjectclass from which all other OpenML objects like
CRAN release: 2021-09-24
CRAN release: 2021-02-22
- If not set via option
mlr3oml.api_key, the API key is retrieved from the environment variable
- Implemented a retry mechanism as a workaround for temporary connection errors.
CRAN release: 2020-10-20
- The parsers for ARFF files can now be explicitly selected via option
"mlr3oml.arff.parser". Default is the internal parser based on
- Improved stability of the internal ARFF parser in case of malformed ARFF files and non-standardized quotes.
CRAN release: 2020-10-05
- The connectors used in
mlr_resamplingsnow signal errors of class
missingDefaultErrorif some defaults are not set.
- Target columns are now automatically converted to the require storage mode during task creation.
- Removed dependency on orphaned package
CRAN release: 2020-08-24