fit_fformpp: Function to fit Efficient Bayesian Surface Regression Models.
In thiyangt/fformpp: Feature-based FORecast Model Performance Prediction

fit_fformpp

R Documentation

Function to fit Efficient Bayesian Surface Regression Models.

Description

This function is written based on the example codes in movingknots package by Dr. Feng Li. Please see https://github.com/feng-li/movingknots for more details.

Usage

fit_fformpp(
  feamat,
  accmat,
  sknots = 2,
  aknots = 2,
  fix.s = 0,
  fix.a = 0,
  fix.shrinkage,
  fix.covariance = 0,
  fix.coefficients = 0,
  n.iter = 100,
  knot.moving.algorithm = "Random-Walk",
  ptype = c("identity", "identity", "identity"),
  prior.knots
)

Arguments

`feamat`	matrix of features, rows corresponds to each time series, columns coresponds to features
`accmat`	matrix of forecast errors from each method, rows represent time series, columns represent forecast algorithms
`sknots`	the dimension of knots for surface, default 2
`aknots`	no. of knots used in each covariates for the additive part, default 2
`fix.s`	number of knots to be fixed in the surface components, 0 means all are updated
`fix.a`	number of knots to be fixed in the additive components, default is 0 which means all are updated
`fix.shrinkage,`	number of shrinkage covariates not to be updated, defalut is, 1:p,
`fix.covariance,`	number of knots to be fixed in the covariance, default is 0, all are updated
`fix.coefficients,`	number of knots to be fixed in the coefficients, default is 0, all are updated
`n.iter`	number of ierations
`knot.moving.algorithm,`	select either "KStepNewton" or "Random-Walk", to fasten the code use Random-Walk
`ptype`	For fixing gprior, This could be c("X'X", "identity", "identity") or c("identity", "identity", "identity")
`prior.knots`	this could be n, log(n) or 1 # to set priors for knots