Update parameters and dependencies of all learners

[be-marc]

Closes #9
Closes #27

• Glmnet
  Updated to glmnet 3.0

• KKNN
  Not sure if y-kernel should be implemented

• LDA

• LogReg

• NaiveBayes

• QDA

• Ranger

• SVM

• Xgboost
  Parameters updated in Update xgboost learner to latest CRAN version #31

• KM

• LM

• CV parameter for internal cross-validation are not included

• Should we include na.action if the options are fail or omit records with missing values? This would add the missing property to the learner but only if na.action is set to omit. Some of the learners in the list support this.

• We could add parameters like offset, contrast and singular.ok to LogReg and LM? But they were not included in mlr. Do we want this?

[pat-s]

Thanks for cleaning up!

[codecov-io]

Codecov Report

Merging #46 into master will decrease coverage by 0.07%.
The diff coverage is 97.46%.

@@            Coverage Diff             @@
##           master      #46      +/-   ##
==========================================
- Coverage   97.61%   97.53%   -0.08%     
==========================================
  Files          18       18              
  Lines         630      691      +61     
==========================================
+ Hits          615      674      +59     
- Misses         15       17       +2

Impacted Files	Coverage Δ
R/LearnerClassifXgboost.R	97.8% <ø> (ø)	⬆️
R/LearnerClassifQDA.R	91.66% <100%> (+1.19%)	⬆️
R/LearnerClassifRanger.R	96.07% <100%> (+0.24%)	⬆️
R/LearnerRegrSVM.R	100% <100%> (ø)	⬆️
R/LearnerClassifSVM.R	100% <100%> (ø)	⬆️
R/LearnerRegrGlmnet.R	96.72% <100%> (+0.35%)	⬆️
R/LearnerClassifNaiveBayes.R	100% <100%> (ø)	⬆️
R/LearnerClassifGlmnet.R	96.92% <100%> (+0.14%)	⬆️
R/LearnerClassifLDA.R	92.3% <100%> (+1.39%)	⬆️
R/LearnerRegrRanger.R	96.15% <100%> (+0.5%)	⬆️
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 543b37b...5a3ade6. Read the comment docs.

[pat-s]

We could add parameters like offset, contrast and singular.ok to LogReg and LM? But they were not included in mlr. Do we want this?

If there is no reason mentioned why they were omitted: Then yes, add them. Most likely they were added recently and just not updated yet in mlr2.

---

Could you please add the changes to NEWS.md? Not every single param ofc, just mentioning all updated learners and maybe version info for bigger changes like glmnet and xgboost.