v0.1.2

Changes

Two new postprocessing methods: isotonic and lasso-estimated quantile regressions

Isotonic quantile regression

• estimate quantiles as linear functions with non-negative coefficients, see A. Lipiecki & B. Uniejewski Isotonic Quantile Regression Averaging for uncertainty quantification of electricity price forecasts for more details
• create an isotonic quantile regression using iQR constructor or call point2quant with method=:iqr

Lasso-estimated quantile regression

• estimate quantiles as linear functions with L1 penalty on the coefficients
• select the optimal regularisation strength with Bayesian Information Criterion
• create a lasso-estimated quantile regression using LassoQR constructor or call point2quant with method=:lassoqr