tune
tune copied to clipboard
tidymodels
Tools for tidy parameter tuning
These take bootstrap resampling and use more sophisticated estimates to estimate and correct for bias. I have a prototype implementation used in [aml4td](https://aml4td.org/chapters/resampling.html#correcting-for-bias) in the `632-estimate` branch. Notes: - It...
It is fairly slow, requires indicators for categorical predictors, and the maintainers don't respond to emails. I've been looking into [GauPro](https://github.com/CollinErickson/GauPro). Pros: - it has kernel functions for categorical tuning...
**Heads up--this PR is not yet functional!** Enables tuning arguments in tailor postprocessors. Works in some cases, but still fails most others. Importantly, this disables the submodel trick right now...
I am working through a two-outcome classification model workflow. My response variable was a two-level factor named `idr_outcome`. I attempted to tune a boosted tree model with cross validation via...
## The problem I'm having trouble fitting the best cross-validated mlp model on my entire training set when the model has multiple hidden layers. This is using the tidymodels framework....
`tune:::create_initial_set()` will use `dials::grid_space_filling()` when an integer is passed. We have space-filling designs for 2+ points and when `grid = 1`, it creates a random grid but uses a LHD...
## Feature suggestion Now that we have the new {tailor} package for post-processing in titydmodels, I find myself in the need to reuse the splits from tune_results as a resampling...
Closes #955. A proof of concept for cutting out duplicated preprocessor fits for iterative tuning approaches. A bit of a contrived example, but on `main`: ``` r library(tidymodels) library(embed) hotel_rates$arrival_date...
## Feature Currently, it appears that `tune_bayes` recomputes the entire preprocessor during every iteration, even if the preprocessor has nothing to tune. This can lead to a substantial amount of...
Hi Max, thanks for keeping vigilant and up to date! If you want to continue to test the labels, we also implemented an accessor (https://github.com/tidyverse/ggplot2/pull/6078) for exactly this reason. _Originally...