Refit function for cv_quantile_genlasso object — refit_quantile

Refit generalized lasso solutions at a new set of quantile levels, given an existing cv_quantile_genlasso object.

refit_quantile_genlasso(
  obj,
  x,
  y,
  d,
  tau_new,
  weights = NULL,
  intercept = NULL,
  standardize = NULL,
  lb = NULL,
  ub = NULL,
  noncross = FALSE,
  x0 = NULL,
  lp_solver = NULL,
  time_limit = NULL,
  warm_starts = NULL,
  params = NULL,
  transform = NULL,
  inv_trans = NULL,
  jitter = NULL,
  verbose = FALSE
)

Arguments

obj: The cv_quantile_genlasso object to start from.
x: Matrix of predictors.
y: Vector of responses.
d: Matrix defining the generalized lasso penalty.
tau_new: Vector of new quantile levels at which to fit new solutions.
noncross: Should noncrossing constraints be applied? These force the estimated quantiles to be properly ordered across all quantile levels being considered. The default is FALSE. If TRUE, then noncrossing constraints are applied to the estimated quantiles at all points specified by the next argument x0. Note: this option only makes sense if the values in the tau vector are distinct, and sorted in increasing order.
x0: Matrix of points used to define the noncrossing constraints. Default is NULL, which means that we consider noncrossing constraints at the training points x.
verbose: Should progress be printed out to the console? Default is FALSE.

Value

A quantile_genlasso object, with solutions at quantile levels

tau_new.

Details

This function simply infers, for each quantile level in tau_new, a (very) roughly-CV-optimal tuning parameter value, then calls quantile_genlasso at the new quantile levels and corresponding tuning parameter values. If not specified, the arguments weights, intercept, standardize, lb, ub, lp_solver, time_limit, warm_starts, params, transform, inv_transorm, jitter are all inherited from the given cv_quantile_genlasso object.