2018/12/18 by Xiyue Liao, Liao, Xiyue, Mary C. Meyer +1
Decision Sciences · Mathematics · #Advanced Statistical Methods and Models #Applications (stat.AP) #FOS: Computer and information sciences #Optimal Experimental Design Methods #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.1812.07696
openalex publication_date 2018/12/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
The cgam package contains routines to fit the generalized additive model where the components may be modeled with shape and smoothness assumptions. The main routine is cgam and nineteen symbolic routines are provided to indicate the relationship between the response and each predictor, which satisfies constraints such as monotonicity, convexity, their combinations, tree, and umbrella orderings. The user may specify constrained splines to fit the components for continuous predictors, and various types of orderings for the ordinal predictors. In addition, the user may specify parametrically modeled covariates. The set over which the likelihood is maximized is a polyhedral convex cone, and a least-squares solution is obtained by projecting the data vector onto the cone. For generalized models, the fit is obtained through iteratively re-weighted cone projections. The cone information criterion is provided and may be used to compare fits for combinations of variables and shapes. In addition, the routine wps implements monotone regression in two dimensions using warped-plane splines, without an additivity assumption. The graphical routine plotpersp will plot an estimated mean surface for a selected pair of predictors, given an object of either cgam or wps. This package is now available from the Comprehensive R Archive Network at http://CRAN.R-project.org/package=cgam.