In mlincon/pvurt: P-value for augmented Dickey-Fuller unit root test

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

pvurt

This package has been created as part of my master thesis. The package has only one user facing function: computePVale(), which approximates p-value (cumulative probability) for augmented Dickey-Fuller unit root test based on the distribution of the test statistics. The approximation is based on pre-trained generalized additive logistic model or logistic regression with polynomial terms.

The function can take either a numeric value or the output from the four functions of the following three packages: adfTest() and unitrootTest() from fUnitRoots, adf.test() from tseries and, lastly, ur.df() from urca. If output from any of the function is passed on, computePVale() appends its approximation result to the original output.

Installation

The package has not been released in CRAN. To install use:

devtools::install_github("mlincon/pvurt")

Since the package must be complied, ensure that Rtools.exe is installed beforehand.

Example

library(pvurt)

y <- arima.sim(model = list(order = c(0, 1, 0)), n = 100)

# Test type: with drift and trend
# package: fUnitRoots
library(fUnitRoots)
computePValue(adfTest(y, lags = 3, type = "ct"))

computePValue(unitrootTest(y, lags = 3, type = "ct"))

# package: urca
library(urca)
computePValue(ur.df(y, lags = 3, type = "trend"))
# print summary
summary(computePValue(ur.df(y, lags = 3, type = "trend")))

# package: tseries
library(tseries)
computePValue(adf.test(y, alternative = "stationary", k = 3))

# no packages
tStat <- -2.239
sampleSize <- 100
computePValue(tStat, n = sampleSize, model = "gam", type = "ct")