workboots
workboots copied to clipboard
markjrieke
Generate bootstrap :hiking_boot: prediction intervals from a tidymodels workflow!
In the paper [Bootstrap Prediction Intervals in Non-Parametric Regression...](https://ntrs.nasa.gov/citations/20130014367) ([Dan Saattrup Nielsen's post on implementation](https://www.saattrupdan.com/posts/2020-03-01-bootstrap-prediction), [my post mostly mirroring Dan's](https://www.bryanshalloway.com/2021/04/05/simulating-prediction-intervals/)) which I believe follows essentially the same procedure as workboots,...
Closes #56 The main API change is that a fitted workflow would now be required. That made the changes a little more invasive than I hoped they would be since...
Looking at the internals, there are some changes you could make to potentially get some speedups. Inside tidymodels, I tend to use tidyverse functions (e.g. `filter()` and `select()`) instead of...
The computations do a straightforward translation of what is in the 632+ paper. A few months ago I emailed Brad Efron about this asking if it was roughly equivalent to...
If we have a formula method,`predict_boots()` fails ``` r library(tidymodels) library(workboots) car_pred_int % predict_boots( n = 2000, training_data = mtcars[3:32,], new_data = mtcars[1:2,] ) #> Error in UseMethod("filter"): no applicable...
Once you’ve got a final model, you might want to make interval predictions multiple times. I think that we can save the bootstrap models (efficiently) so that we don’t have...
Right now, workboots doesn't do anything special in terms of parallelization for speed improvements. Replacing the `purrr::map()` functions with `furrr::future_map()` functions under the hood doesn't actually speed things up (I...
see [this dplyr article](https://dplyr.tidyverse.org/articles/programming.html#eliminating-r-cmd-check-notes) on using the `.data` pronoun from tidyr
see here: https://dplyr.tidyverse.org/articles/programming.html#eliminating-r-cmd-check-notes
Metadata
Owner
Metadata
Generate bootstrap :hiking_boot: prediction intervals from a tidymodels workflow!