pipeflow (version 0.2.2)

pipe_set_data_split: Split-multiply pipeline by list of data sets

Description

This function can be used to apply the pipeline repeatedly to various data sets. For this, the pipeline split-copies itself by the list of given data sets. Each sub-pipeline will have one of the data sets set as input data. The step names of the sub-pipelines will be the original step names plus the name of the data set.

Usage

pipe_set_data_split(
  pip,
  dataList,
  toStep = character(),
  groupBySplit = TRUE,
  sep = "."
)

Value

new combined Pipeline with each sub-pipeline having set one of the data sets.

Arguments

pip

Pipeline object

dataList

list of data sets

toStep

string step name marking optional subset of the pipeline, to which the data split should be applied to.

groupBySplit

logical whether to set step groups according to data split.

sep

string separator to be used between step name and data set name when creating the new step names.

Examples

# Split by three data sets
dataList <- list(a = 1, b = 2, c = 3)
p <- pipe_new("pipe")
pipe_add(p, "add1", \(x = ~data) x + 1, keepOut = TRUE)
pipe_add(p, "mult", \(x = ~data, y = ~add1) x * y, keepOut = TRUE)
pipe_set_data_split(p, dataList)
p

p |> pipe_run() |> pipe_collect_out() |> str()

# Don't group output by split
p <- pipe_new("pipe")
pipe_add(p, "add1", \(x = ~data) x + 1, keepOut = TRUE)
pipe_add(p, "mult", \(x = ~data, y = ~add1) x * y, keepOut = TRUE)
pipe_set_data_split(p, dataList, groupBySplit = FALSE)
p

p |> pipe_run() |> pipe_collect_out() |> str()

# Split up to certain step
p <- pipe_new("pipe")
pipe_add(p, "add1", \(x = ~data) x + 1)
pipe_add(p, "mult", \(x = ~data, y = ~add1) x * y)
pipe_add(p, "average_result", \(x = ~mult) mean(unlist(x)), keepOut = TRUE)
p
pipe_get_depends(p)[["average_result"]]

pipe_set_data_split(p, dataList, toStep = "mult")
p
pipe_get_depends(p)[["average_result"]]

p |> pipe_run() |> pipe_collect_out() |> str()