mlr3pipelines

mlr3pipelines copied to clipboard

`PipeOpEncode` doesn't work for features of type `character()`

2

The [doc](https://mlr3pipelines.mlr-org.com/reference/mlr_pipeops_encode.html) says that is should encode `character()` columns, but: ``` r library(mlr3) library(mlr3pipelines) data = data.table::data.table(x = letters[1:3], y = factor(letters[1:3])) task = as_task_classif(data, id = "task", target =...

bblodfon

`mlr3` has its test helpers in `inst/testthat/` so other packages can load it and use them (https://github.com/mlr-org/mlr3/tree/main/inst/testthat) It would be convenient if the mlr3pipelines test helpers also lived in `inst`

sebffischer

crate all custom_check functions

1

sebffischer

https://github.com/mlr-org/mlr3pipelines/blob/dae03afd8efa83621a495a058f42cfc2ea8d7357/R/ppl.R#L6 Replace `charcter(1)` with `character(1)`.

jpoehl

mb706

sebffischer

This is already fixed in this PR: https://github.com/mlr-org/mlr3pipelines/pull/737

sebffischer

check that we are using encapsulate for all pipeop (default) arguments and check functions

mb706

``` r library(mlr3proba) #> Loading required package: mlr3 task = tsk('lung') trgs = task$data(cols = task$target_names) trgs #> time status #> 1: 306 TRUE #> 2: 455 TRUE #> 3:...

bblodfon

In many cases the objects in a graph have some kind of in-place option, and in fact we often explicitly clone objects to avoid this. It is necessary in general,...

mb706