Changes in version 2025-07-13                      

Changed

  - Breaking extensive refactor to literanger object in R; any objects
    serialized with versions < 0.2.0 cannot be deserialized with new
    release.
  - Training speed improved via inlining ca46ed46 a3177475, and
    short-circuits 640912f7 43d26855; mileage may vary but 10-20% faster
    is expected.

Added

  - A merge function to merge trained forests by copying trees.

                     Changes in version 2024-09-13                      

Performance improved. R interface and C++ core have been separated.

Changed

  - Set Depends to R >= 3.6.0.
  - Training speed was increased by ~25% by reducing memory allocations
    19e7c475 and inlining access to data 233063e0.
  - Source code underwent a minor re-organisation to separate the
    R-specific components from the C++ library.

Added

  - DataVector class to read, write, and pass data without R.

                     Changes in version 2024-09-03                      

New feature! literanger can now serialize trained random forests using
cereal.

The project has been moved to GitLab:
https://gitlab.com/stephematician/literanger.

Changed

  - The value-type returned by predict now matches the response type in
    training ea67c83e
  - Bump cpp11 to 0.4.7.

Added

  - Functions read_literanger and write_literanger for serialization.

Fixed

  - Fixed bug in implementation of always-selected candidates for
    splitting, e.g. the names_of_always_draw argument 6d31d7f3.
  - Minor performance tweak 9a3b639a in particular for 'maxstat'
    37580d9b.

                     Changes in version 2023-07-11                      

Update to pass CRAN's ASAN check

Changed

  - Improve performance of node splitting d3f6424.

Added

  - Add re-entrant log gamma to speed up beta splitting rule d7f058d.
  - Minor fixes to documentation 91b6c6d 0f62d02.

Fixed

  - Fix potential illegal access and incorrect unweighted sampling
    without replacement b6df5d9.

                     Changes in version 2023-06-25                      

First release

A refactoring and adaptation of the ranger package
https://github.com/imbs-hl/ranger for random forests. Has faster
prediction mode intended for embedding into the multiple imputation
algorithm proposed by Doove et al in:

Doove, L. L., Van Buuren, S., & Dusseldorp, E. (2014). Recursive
partitioning for missing data imputation in the presence of interaction
effects. Computational statistics & data analysis, 72, 92-104.

Added

  - Fit classification and regression trees
  - Prediction via most frequent value or mean
  - Get predictions as terminal node identifiers in each tree or as a
    random draw from inbag values in a random tree