mlr3pipelines

mlr3pipelines copied to clipboard

Is there a way to transform a new data as per pipeline trained on a given task. For example:- ```r ames = mlr3data::ames_housing ames1 = ames[1:2000,] ames2 = ames[2001:nrow(ames),-c('Sale_Price')] to_remove...

mermast

E.g.: make `.result` part of the `$state`? -- has some drawbacks and does not work for predictions, and may be incompatible with states that are "closed" (i.e. pipeop does not...

mb706

mb706

... since using `ppl()` already makes it possible to change them. Instead of doing `ppl("bagging", frac = 0.5)`, one can already do `ppl("bagging", subsample.frac = 0.5)`, because of the way...

mb706

mb706

PipeOpThreshold has wrong predict type

3

PipeOpThreshold claims to have `predict_type` `"response"` but actually outputs probabilities. Also, why is it not possible to change the predict type of the `PipeOp`?

sebffischer

Better warning if po(<id>) does not work, but ppl(<id>) exists

3

Inspired by this SO post: https://stackoverflow.com/users/17945841/programming-noob

sebffischer

pos in combination with as_graph leads to unexpected ids

1

``` r library(mlr3pipelines) as_graph(pos(c("pca", "ica"))) #> Graph with 2 PipeOps: #> ID State sccssors prdcssors #> pca.pca #> ica.ica ``` Created on 2023-04-20 with [reprex v2.0.2](https://reprex.tidyverse.org)

sebffischer

filter hash not entirely correct

1

depends on https://github.com/mlr-org/mlr3filters/issues/154

mb706

sebffischer