stablelearner 0.1-8
Code repository changed from R-Forge to Codeberg at: https://codeberg.org/zeileis/stablelearner/
Added basic altdoc page with overview and documentation at: https://zeileis.codeberg.page/stablelearner/
stablelearner 0.1-7
- Fix small bug in
splithalf()sampling forstability(). In case of an uneven number of observations in the learning sample some small overlap in the two splitted halves would occur. This is avoided now by dropping one random observation now and restricting both halves to be always of the same size. Reported by Constantin Wiegand along with two further small improvements: Warnings about lack of variance in predictions on learning samples and avoiding manual triggering of the garbage collector.
stablelearner 0.1-6
- Improve non-anchored links in manual pages (prompted by CRAN).
stablelearner 0.1-5
The
as.stabletree()method forRandomForestobjects (party package) is now registered as an S3 (rather than S4) method.Environment
.stabEnvis only used internally bystabletree()and hence not exported anymore inNAMESPACE.
stablelearner 0.1-4
Z takes over maintenance from MP.
Fix checks for
samplerspecifications where a logical of length 2 is now correctly aggregated withall().
stablelearner 0.1-3
- Small improvements for CRAN checks.
stablelearner 0.1-2
Changed default sampling method in
stabletree()frombootstrap()tosubsampling()with default fraction ofv = 0.632.as.stabletree()coercion generic added which allows to coerce arandomForest(randomForest package),RandomForest(party package),cforest(partykit package) orranger(ranger package) to astabletreeobject.Added a vignette on the variable and cutpoint selection analysis of random forests.
stablelearner 0.1-1
Project
stablelearnerhas been launched and a stable version of the package has been uploaded to CRAN.stability()is available to estimate the stability of the results from a given supervised statistical learning method.stabletree()is available to estimate the stability of the results from recursive partitioning.