Predict function for quantile_genlasso_grid object

Source: R/quantile_genlasso.R

Predict the conditional quantiles at a new set of predictor variables, using the generalized lasso coefficients at given tau or lambda values.

# S3 method for quantile_genlasso_grid
predict(
  object,
  newx,
  sort = FALSE,
  iso = FALSE,
  nonneg = FALSE,
  round = FALSE,
  ...
)

Details

This function operates as in the predict.quantile_genlasso function for a quantile_genlasso object, but with a few key differences. First, the output is reformatted so that it is an array of dimension (number of prediction points) x (number of tuning parameter values) x (number of quantile levels). This output is generated from the full set of tau and lambda pairs stored in the given quantile_genlasso_grid object obj (selecting a subset is disallowed). Second, the arguments sort and iso operate on the appropriate slices of this array: for a fixed lambda value, we sort or run isotonic regression across all tau values.