simulateUnivariateAR: Builds an autoregressive model and uses it to predict
In olshena/COVIDNearTerm: Near Term Prediction of COVID-Related Population Metrics

View source: R/simulateUnivariateAR.R

simulateUnivariateAR

R Documentation

Builds an autoregressive model and uses it to predict

Description

A wrapper function for buildAR and predictAR.

Usage

simulateUnivariateAR(
  vec,
  x = NULL,
  x_lag = 0,
  wsize = 14,
  method = c("equal", "unweighted", "triangle"),
  pdays = 28,
  nsim = 100,
  skip = 0,
  seed = NULL,
  output_type = "all",
  rhat_method = c("none", "geometric", "arithmetic"),
  lambda = seq(0, 1, 0.05),
  alpha = 0,
  rolling_mean = 1,
  debug = FALSE
)

Arguments

`vec`	A vector of numeric data
`x`	A vector of numeric data, used to predict vec. Must be the same size as vec and indexed to vec
`x_lag`	An integer used to specify the number of observations x should be shifted. Vec will also be truncated on top by x_lag
`wsize`	Number of prior observations to use for averaging, default is 14
`method`	Type of weighting to use, default is equal
`pdays`	Number of days into the future to make predictions, default is 28
`nsim`	Number of simulations, default is 100
`skip`	Number of input values to skip, default is 0
`seed`	Seed for random number generator
`output_type`	Type of output, default is all
`rhat_method`	Method for calculating rhat, if "none", rhat = 1 and has no effect
`lambda`	Shrinkage parameter, if not specified, default is a grid search from 0 to 1 by 0.05. A value of 0 produces no shrinkage. If an array is specified, all values in the array are evaluated and the optimal lambda is chosen based on residual sum of squares. Values should be between 0 and 1 inclusive.
`alpha`	Alpha parameter, if not specified, default is 0 which produces no scaling. If an array is specified, all values in the array are evaluated and the optimal alpha is chosen based on residual sum of squares. Values should be >=0.
`rolling_mean`	rolling mean window for x. Must be an integer less than length of x and at least 1. Default is 1
`debug`	TRUE returns buildAR objects in addition to standard output