mlr3viz is the visualization package of themlr3 ecosystem. It features plots for mlr3 objects such as tasks, learners, predictions, benchmark results, tuning instances and filters via theautoplot() generic ofggplot2. The package draws plots with theviridis color palette and applies theminimal theme. Visualizations include barplots, boxplots, histograms, ROC curves, and Precision-Recall curves.
Thegallery features ashowcase post of the plots inmlr3viz.
Installation
Install the last release from CRAN:
install.packages("mlr3")Install the development version from GitHub:
remotes::install_github("mlr-org/mlr3viz")Resources
Thegallery features a showcase post of the visualization functionsmlr3viz.
Short Demo
library(mlr3)library(mlr3viz)task=tsk("pima")learner=lrn("classif.rpart", predict_type="prob")rr=resample(task,learner,rsmp("cv", folds=3), store_models=TRUE)# Default plot for taskautoplot(task, type="target")
# ROC curve for resample resultautoplot(rr, type="roc")
For more example plots you can have a look at thepkgdown references of the respective functions.