mlr3shiny
mlr3shiny copied to clipboard
LamaTe
mlr3shiny: Machine Learning in Shiny with mlr3
#67 -added input for custom label support in learner-tab -label will be shown instead of id #71 -added progess bar for benchmark scoring computation
...for multiclass problems if predict type is response and not prob. (The automatic error message is hard to understand for users non-famila with mlr3.)
...to predict tab (between train learner and export learner): Use print(learner$model) for some basic information on the trained learner. Check for th different available learners: Does this fit into the...
- maintainability aspects - additonal window below the task definition - is there space for the plots? - ...type = "pair" (classif & regr) vs. "duo" (only classif) - add...
repex: - [task: iris -> no change] - learner: xgboost -> change parameter eta to e.g. 0.2 -> click change parameters Warning: Error in if: argument is of length zero...
(...as this sometimes same a couple of time) additional minor issue: btw. it sometimes there is a notable time between completion of the benchmark progress bar and the occurrence of...
REPEX: task = gcd + deselect robustification -> learner = svm + xgboost -> train learner -> error: training failed Add 'Try robustificationin Task tab' to message.
repex: task iris select features -> uncheck petal.width, petal.length click select click select again -> features disappear from list and cannont be re-selected (unless a new task is created) ...repeat...
Check whether PMML is still up to date and whether R package pmml allows for an easy implementation. Could be added to the predict tab.