predictModel: Use results from function estimateModel to make predictions.

In StatisticsNZ/demest: Bayesian Demographic Estimation and Forecasting

Use results from function estimateModel to make predictions.

Description

Typically predict along time dimension, but can predict along other dimenions. When predicting along time dimension, typically predict forwards, but can predict backwards.

Usage

predictModel(
  filenameEst,
  filenamePred,
  along = NULL,
  labels = NULL,
  n = NULL,
  exposure = NULL,
  weights = NULL,
  data = NULL,
  aggregate = NULL,
  lower = NULL,
  upper = NULL,
  nBurnin = 0L,
  parallel = TRUE,
  outfile = NULL,
  verbose = FALSE,
  useC = TRUE
)

Arguments

filenameEst	Filename used to estimateModel.
filenamePred	Filename to be used by predictModel.
along	Name or index along which prediction should occur.
labels	Labels of new categories.
n	Number of new categories. Can only be used when predicting along "time", "age", or "cohort" dimensions, and when the units have equal length.
exposure	A Counts object specifying exposure or sample size.
weights	A Counts object containing weights.
data	A named list containing future values of covariates.
aggregate	An object of class SpecAggregate.
lower	A lower bound for estimates of data-level rates, probabilities, or means.
upper	An upper bound for estimates of data-level rate, probabilities, or means.
nBurnin	Number of iteration discarded before recording begins.
parallel	Logical. If TRUE (the default), parallel processing is used.
outfile	Where to direct the ‘stdout’ and ‘stderr’ connection output from the workers when parallel processing. Passed to function [parallel]{makeCluster}.
verbose	Logical. If TRUE (the default) a message is printed at the end of the calculations.
useC	Logical. If TRUE (the default), the calculations are done in C. Setting useC to FALSE may be useful for debugging.

Details

If an along argument is not supplied, predictModel looks for a dimension with dimtype "time", and, failing that, a dimension with dimtype "age", or "cohort".

When predicting along "time", "age" or "cohort" dimensions, specifying an n argument is usually more convenient than specifying a labels argument.

exposure or weights arguments are needed if predictions for outcome variable y, and not just the model parameters, are needed.

StatisticsNZ/demest documentation built on Nov. 2, 2023, 7:56 p.m.