Fit and forecast a time series using RVFL

This function combines model fitting using tsrvfl() and forecasting in one step. It's a convenience wrapper that fits a model and immediately generates forecasts.

tsrvflf(
  y,
  h = 5,
  lags = 15L,
  engine = stats::lm,
  level = 95,
  n_hidden_features = 5L,
  nodes_sim = c("sobol", "halton", "unif"),
  activ = c("relu", "sigmoid", "tanh", "leakyrelu", "elu", "linear"),
  type_pi = c("surrogate", "kde", "bootstrap"),
  B = 250L,
  agg = c("mean", "median"),
  seed = 123,
  coeffs = NULL,
  ...
)

Arguments

y

A numeric vector or time series of the response variable.

h

The number of steps ahead to forecast.

lags

The number of lags to use in the model.

engine

The engine to use for the model fitting.

level

The confidence level for the prediction intervals.

n_hidden_features

The number of hidden features to use in the model.

nodes_sim

The method to use for simulating nodes.

activ

The activation function to use in the model.

type_pi

The type of prediction interval to use.

B

The number of bootstrap samples to use.

agg

The aggregation function to use.

seed

The random seed to use.

coeffs

The coefficients to use in the model.

Value

An object of class "forecast" containing the fitted model and forecasts

Examples

y <- ts(rnorm(120,0,3) + 1:120 + 20*sin(2*pi*(1:120)/12), frequency=12)
fit <- tsrvflf(y, h=20)
#> Error in tsrvflf(y, h = 20): could not find function "tsrvflf"
plot(fit)
#> Error: object 'fit' not found