TS_adjust_gamma_rho: Adjustments for $gamma$ and $$rho$.

Description Usage Arguments Value

View source: R/TS_adjust_gamma_rho.R

Description

The estimation formulas used for autocovariances and autocorrelations at lag $h$ are connected to the estimation formulas of the covariances and correlation of the corresponding lagged pairs. This function computes the scaling and adjustments components that we need in other functions later on.

Usage

1
2
TS_adjust_gamma_rho(TS, lag_max = NULL, type = c("rho", "gamma"),
  .adjustment_rule = "data", all_details = FALSE)

Arguments

TS

The time series object that we want to work upon. It's assumed that this has been converted into an array, with one dimension-name equal to "content".

lag_max

Integer that decides the number of scaling and adjustment-terms that will be returned. The default value NULL will imply that these terms are included from lag 0 to length(TS)-2. (In practise this will be way above what will be used in the estimates, so this argument should be specified in order to avoid wasting computational resources.)

type

One of c("rho", "gamma"), i.e. should the adjustment-terms be computed for data based on local Gaussian correlations or local Gaussian covariances. If no selections are made, both alternatives will be computed.

.adjustment_rule

Either a non-negative number, or "data". This will be added as an attribute to the normalised time series, and later on it will decide if any finite-sample adjustment should be used for the estimated local Gaussian autocorrelations. Note that no adjustments will be performed when .adjustment_rule=0.

all_details

Logic value, default FALSE, that influences the amount of information returned to the workflow.

Value

An array will be returned giving the scaling and adjustment terms required to go from estimates covariances and correlations to the estimates of autocovariances and autocorrelations. If all_details is TRUE, then some intermediate estimates will be included in the result.


LAJordanger/localgaussSpec documentation built on July 28, 2017, 12:15 a.m.