growlen: Expand Data Length

In nelson16silva/wavsigmap: Wavelet-Based Signal Estimation

Expand Data Length

Description

growlen exapands the length of a time series or a numeric vector by running an auto arima regression for forecasting and backcasting. Extending the sample by inserting forecast can be a simple remedy for the end point problem.

Usage

growlen(x, h, direction = one_side)

Arguments

x	A time series or a numeric vector.
h	An integer. Number of periods for expanding the data.
direction	Where to grow the data. Allowed values are one_side (default) and both. The one_side grows the data on the end of the series (forecasting). both grows on both sides of the series (auto arima backcasting and forecasting).

Value

A time series or a numeric vector. The length of the returned value is the original length plus h.

Examples

GDPC1_grow <- growlen(GDPC1, h = 4)

growlen(GDPC1, h = 8, direction = both)

# EbayesThresh

GDPC1_grow_wavelet <- signal(GDPC1_grow, boundary = "reflection", n.levels = 5, vscale = "level")

gap_wavelet <- 100 * (GDPC1_grow - GDPC1_grow_wavelet) /  GDPC1_grow_wavelet

# Comparing gap_wavelet (end point adjust) with the original GDPC1_GDPPOT

ts.plot(ts.intersect(GDPC1_GDPPOT, gap_wavelet), col = c(1, 2)); abline(h = 0)
legend(1980, 9, legend = c("GDPC1_GDPPOT", "gap_wavelet"), col = c(1, 2), lty = 1)

nelson16silva/wavsigmap documentation built on March 7, 2023, 10:45 a.m.