Active Set and Generalized PAVA for Isotone Optimization
Active Set Methods for Isotone Optimization
Asymmetric Least Squares
Absolute Value Norm
L1 approximation
User-Specified Loss Function
Generalized Pooled-Adjacent-Violators Algorithm (PAVA)
Huber Loss Function
SILF Loss
General Least Squares Loss Function
Least Squares Loss Function
Regression with Linear Inequality Restrictions on Predicted Values
Chebyshev norm
Lp norm
Quantile Regression
Negative Poisson Log-Likelihood
Weighted Median
Weighted Median
Contains two main functions: one for solving general isotone regression problems using the pool-adjacent-violators algorithm (PAVA); another one provides a framework for active set methods for isotone optimization problems with arbitrary order restrictions. Various types of loss functions are prespecified.