Posterior probability calibration methods for machine learning algorithms in R. It provides calibration methods such as isotonic regression, knn-based methods, Bayesian Binning into Quantiles (BBQ), Ensemble of Linear Trend Estimation (ELiTe) in multiclass case. It also provides utilities for calibration performance measures and reliability diagrams.
To install this R package, please run the following code in R
devtools::install_github("kenneth-lee-ch/calibtools")
Then, please see this example for learning how the package can be used.
Please refer to the paper "A Closer Look at Probability Calibration in Multiclass Problems" for details.