In parsnip: A Common API to Modeling and Analysis Functions

#| child: aaa.Rmd
#| include: false

r descr_models("mlp", "qrnn")

Tuning Parameters

#| label: qrnn-param-info
#| echo: false
defaults <- 
  tibble::tibble(parsnip = c("hidden_units", "penalty", "epochs","activation"),
                 default = c("2L", "0.0", "5000L", "'sigmoid'"))

spec <-
  mlp(
    hidden_units = tune(),
    penalty = tune(),
    epochs = tune(),
    activation = tune()
  ) |> 
  set_engine("qrnn")

param <- spec |> make_parameter_list(defaults)

This model has r nrow(param) tuning parameters:

#| label: qrnn-param-list
#| echo: false
#| results: asis
param$item

Other engine arguments of interest:

• n.trials: number of repeated trials used to avoid local minima.
• method: The optimization technique ("nlm" or "adam").

Translation from parsnip to the original package (quantile regression)

#| label: qrnn-qnt
mlp(
  hidden_units = integer(1),
  penalty = double(1),
  epochs = integer(1),
  activation = character(1)
) |>  
  set_engine("qrnn") |> 
  set_mode("quantile regression", quantile_levels = (1:3) / 4) |> 
  translate()

Preprocessing requirements

#| child: template-makes-dummies.Rmd

#| child: template-same-scale.Rmd

Case weights

#| child: template-no-case-weights.Rmd

Prediction types

#| label: predict-types

parsnip:::get_from_env("mlp_predict") |>
  dplyr::filter(engine == "qrnn") |>
  dplyr::select(mode, type)

References

• Cannon, A.J., 2018. Non-crossing nonlinear regression quantiles by monotone composite quantile regression neural network, with application to rainfall extremes. Stochastic Environmental Research and Risk Assessment, 32(11): 3207-3225. doi:10.1007/s00477-018-1573-6

parsnip documentation built on May 14, 2026, 5:08 p.m.