Skip to contents

The progressify package allows you to easily add progress reporting to sequential and parallel map-reduce code by piping to the progressify() function. Easy!

TL;DR

library(progressify)
handlers(global = TRUE)
library(sandwich)

fit <- lm(dist ~ speed, data = cars)
v <- vcovBS(fit, R = 100L) |> progressify()

Introduction

This vignette demonstrates how to use this approach to add progress reporting to sandwich functions such as vcovBS() and vcovJK().

The sandwich package provides model-robust standard error estimators for cross-section, time series, and longitudinal data. Some of these estimators, specifically the bootstrap and jackknife estimators, are computationally intensive and can benefit from progress reporting.

For example, vcovBS() computes bootstrapped covariance matrix estimators.

library(sandwich)
fit <- lm(dist ~ speed, data = cars)
v <- vcovBS(fit, R = 100L)

Here vcovBS() provides no feedback on how far it has progressed, but we can easily add progress reporting by using:

library(sandwich)

library(progressify)
handlers(global = TRUE)

fit <- lm(dist ~ speed, data = cars)
v <- vcovBS(fit, R = 100L) |> progressify()

Similarly, the jackknife estimator vcovJK() can be progressified:

library(sandwich)

library(progressify)
handlers(global = TRUE)

fit <- lm(dist ~ speed, data = cars)
v <- vcovJK(fit) |> progressify()

Supported Functions

The progressify() function supports the following sandwich functions:

Limitations

The progressify() function cannot yet be combined with futurize() of the futurize package for sandwich functions, because both work by setting the applyfun argument, which only one of them can control. With |> progressify() |> futurize(), there is an error. With |> futurize() |> progressify(), there is progress reporting, but the processing is sequential.