Progress updates for 'partykit' functions
Henrik Bengtsson
Source:vignettes/progressify-81-partykit.md
progressify-81-partykit.RmdThe 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(partykit)
data("Titanic", package = "datasets")
tt <- as.data.frame(Titanic)
forest <- cforest(Survived ~ ., data = tt, ntree = 50L) |> progressify()Introduction
This vignette demonstrates how to use this approach to add progress
reporting to partykit
functions such as cforest().
The partykit cforest() function is an
implementation of random forests. For example,
library(partykit)
data("Titanic", package = "datasets")
tt <- as.data.frame(Titanic)
forest <- cforest(Survived ~ ., data = tt, ntree = 50L)Here cforest() provides no feedback on how far it has
progressed, but we can easily add progress reporting by using:
library(partykit)
library(progressify)
handlers(global = TRUE)
data("Titanic", package = "datasets")
tt <- as.data.frame(Titanic)
forest <- cforest(Survived ~ ., data = tt, ntree = 50L) |> progressify()Using the default progress handler, the progress reporting will appear as:
|===== | 20%
Similarly, we can report on progress when calculating the variable importance of the trees in the forest:
vi <- varimp(forest) |> progressify()Supported Functions
The progressify() function supports the following
partykit functions:
Limitations
The progressify() function cannot yet be combined with
futurize() of the futurize package
for partykit 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.