fit_kernel_smoothed_stacked_model_fixed_bw: Estimate weights for a linear combination of predictive...

In reichlab/adaptively-weighted-ensemble: Adaptively Weighted Ensembles via Stacking

Description

This function uses the "degenerate EM" algorithm outlined at http://www.cs.cmu.edu/~roni/11761-f16/Presentations/degenerateEM.pdf, but modified in a way that is ad hoc but I think could be justified to use kernel-weighted observations.

Usage

fit_kernel_smoothed_stacked_model_fixed_bw(component_model_log_scores,
  covariate, prediction_covariate, bw, tol = .Machine$double.eps)

Arguments

component_model_log_scores	a data frame or matrix of log scores. Each column gives log scores for a particular predictive model. Each row corresponds to one observation
covariate	a single covariate over which weights should be smoothed (more than one covariate may be supported later)
bw	a bandwidth for the smoothing
tol	numeric, if method was "em", stopping tolerance

Value

model weights

reichlab/adaptively-weighted-ensemble documentation built on May 27, 2019, 4:53 a.m.