sigex.fixed: Computes all nontrend fixed regression effects

View source: R/sigex.fixed.r

sigex.fixedR Documentation

Computes all nontrend fixed regression effects

Description

Background: x is a multivariate time series (N x T), and each individual series can have its distinct set of regressors. So for each 1 <= j <= N, x[j,] is length T and has r_j number of length T regressors. There is a default regressor of polynomial time: suppose the time series has d unit roots (d >= 0), and this applies to each individual series (differencing polynomials are the same for all individual series in sigex). Then the regressor t^d for 1 <= t <= T is the default "mean effect". (Coefficients of lower order time polynomial effects cannot be identified.) When d=0, this is just the mean of the process. (Although it need not be stationary when d=0, any other non-stationary latent components are assumed to have mean zero for identifiability.) One can always add higher order time polynomial regressors, if desired. param is the name for the model parameters entered into a list object with a more intuitive structure, whereas psi refers to a vector of real numbers containing all hyper-parameters (i.e., reals mapped bijectively to the parameter manifold)

Usage

sigex.fixed(data.ts, mdl, series, param, type)

Arguments

data.ts

A T x N matrix ts object

mdl

The specified sigex model, a list object

series

Integer between 1 and N, the index of the individual series for which regression effects are being computed.

param

The model parameters entered into a list object

type

A string designating the name of the regression effect

Value

mean.mat: length T time series consisting of regression effects. This has the format X X is a regressor of "type", and beta consists of the corresponding regression parameters.


jlivsey/sigex documentation built on May 25, 2024, 4:17 a.m.