Add a "Create your own score class object" page for filtro #117

franceslinyc · 2025-08-12T22:03:46Z

This PR adds a new post titled "Create your own score class object" for filtro in developer tools.

Finalize
Have Hannah review once finalizing
Revise based on Hannah's feedback
Ready to merge

franceslinyc · 2025-08-18T18:15:08Z

Error:
! Can't find method for `fit(<class_score_aov>)`.
Hide Traceback
    ▆
 1. ├─filtro::fit(score_aov_pval, Sale_Price ~ ., data = ames_subset)
 2. │ └─S7::S7_dispatch()
 3. └─S7:::method_lookup_error("fit", `<named list>`)

I think this is related to PR #162 in filtro

franceslinyc · 2025-08-19T15:24:40Z

Do users need to install S7?

franceslinyc · 2025-08-19T17:40:48Z

@topepo @EmilHvitfeldt Just want to know if there is anything that is missing from the page.

Still trying to see why fit() throws error here. It won't let me Preview nor Render so I commented it out currently.

franceslinyc · 2025-08-19T17:46:17Z

Error in `UseMethod()`:
! no applicable method for 'fit' applied to an object of class "c('class_score_aov', 'filtro::class_score', 'S7_object')"
Hide Traceback
    ▆
 1. └─generics::fit(score_aov_pval, Sale_Price ~ ., data = ames_subset)

learn/develop/filtro/index.qmd

EmilHvitfeldt · 2025-08-20T03:06:20Z

Do users need to install S7?

Nope, it should be installed as they install {filtro}, as it is in Imports

hfrick

Looks like you made a very cool package with filtro! I would add a little bit more context in the article.

You don't have to write a novel about what filtro does (that's what the filtro website is for) but just a few sentences to situate a reader who's possibly coming from the developer category on tidymodels.org and not directly from the filtro page.

I'd also add some broad strokes about how filtro works as context for the concrete steps of what to actually do when implementing a new scoring method. And then always show concrete examples, like for the docs.

hfrick · 2025-08-21T16:46:36Z

learn/develop/filtro/index.qmd

+
+`r article_req_pkgs(pkgs)`
+
+You can construct new scoring objects using `class_score()`. This article is a guide to creating new scoring objects. 


Can you give readers some context for scoring objects, e.g., via a brief intro to filtro? See the other articles in the "developer tools" category for examples: https://www.tidymodels.org/learn/#category=developer%20tools

hfrick · 2025-08-21T16:50:24Z

learn/develop/filtro/index.qmd

+
+## Scoring object
+
+All subclasses specific to the scoring method have a parent class named `class_score`. There are a few properties (attributes) for this object:


Can you add a little bit more context on the package structure?

hfrick · 2025-08-21T16:56:09Z

learn/develop/filtro/index.qmd

+args(class_score)
+```
+
+-   `outcome_type`: What types of outcome can the method handle? The options are `numeric`, `factor`, or both. 


This is a pretty long list. Does the reader need to know about all of them, at this point? Or could you highlight some and point them to the docs for the rest?

hfrick · 2025-08-21T16:59:00Z

learn/develop/filtro/index.qmd

+
+## Scoring object specific to the scoring method
+
+We demonstrate how to create a custom scoring object specific to the given scoring method.


Does this mean that a scoring object is the implementation of a scoring method? The use of "specific to" is throwing me a little here.

hfrick · 2025-08-21T17:00:42Z

learn/develop/filtro/index.qmd

+
+We demonstrate how to create a custom scoring object specific to the given scoring method.
+
+As an example, let’s consider the ANOVA F-test filter. 


A sentence on the stats aspect of this would be a nice bonus.

hfrick · 2025-08-21T17:05:18Z

learn/develop/filtro/index.qmd

+  )
+```
+
+## Fitting (or estimating) score


Does this use the fit method for the parent class? I was expecting to see some class-specific method implementations in this section.

hfrick · 2025-08-21T17:08:00Z

learn/develop/filtro/index.qmd

+
+Documentation for S7 methods is still a work in progress, and it seems no one currently knows the right approach. Here’s how we tackle it: 
+
+We re-export the `fit()` generic from another package. Instead of documenting each individual `fit()` method, we provide the details in the "Details" section and the "Estimating the scores" subsection of the documentation for the corresponding object. 


I would show examples of that. The call to the help pages below shows users how to access it and that's helpful context but it less focused on "how to implement your own scoring class".

franceslinyc · 2025-08-21T19:50:25Z

All super important points, thanks @hfrick! I'll revise according.

hfrick

Thank you for the added context! I have a much better orientation in this now. 🐸

My understanding is that the core things to know about adding your own scoring method are:

There is a base class for scoring objects. You add a scoring method by

Creating a subclass of that base class
Writing a fit() method for the subclass

And for the docs, you get guidance on how to document an S7 method.

Assuming I got that correct, I think the article could make that just a little clearer and then we're there!

hfrick · 2025-08-29T10:24:37Z

learn/develop/filtro/index.qmd

+
+filtro is tidy tools to apply filter-based supervised feature selection methods. It provides functions to rank and select a specified proportion or a fixed number of features using built-in methods and the desirability function. 
+
+Currently, there are 6 filters in filtro and many existing score objects. A list of existing scoring objects [can be found here](https://filtro.tidymodels.org/articles/filtro.html#available-score-objects-and-filter-methods). However, you might need to define your own scoring objects. This article serves as a guide to creating new scoring objects and computing feature scores before performing ranking and selection. 


What's the purpose of a scoring object? So far, the text only mentions filters.

Ah, some explanation coming at the end of the paragraph. Could we reshuffle that a little?

Just saw these now, adding (revising)! Thanks!

hfrick · 2025-08-29T10:35:08Z

learn/develop/filtro/index.qmd

+
+The general procedure is to:
+
+1. Create a parent scoring object `class_score`, specifying fixed properties that are shared across all custom scoring objects. 


This class is already implemented in filtro, right? If so, we should make it clearer that this is the infrastructure we build on when we make our own score class object.

hfrick · 2025-08-29T10:40:24Z

learn/develop/filtro/index.qmd

+
+1. Create a parent scoring object `class_score`, specifying fixed properties that are shared across all custom scoring objects. 
+
+2. Construct a custom scoring object `class_score_*`, adding additional, implementation-specific properties. 


I'd rephrase this to say that we are making a subclass, inheriting from class_score. The info that this is the place to specify aspects specific to this score (as opposed to the general characteristics already covered by the parent class) is proabaly helpful for orientating readers less familar with object-oriented programming.

hfrick · 2025-08-29T10:43:04Z

learn/develop/filtro/index.qmd

+
+2. Construct a custom scoring object `class_score_*`, adding additional, implementation-specific properties. 
+
+3. Define the scoring method in `fit()`, which computes feature score. `fit()` refers to the custom scoring object from step 2 to use the appropriate method.


If you think it's helpful, you could clarify that defining the subclass corresponds to defining characteristics of your score and writing the fit() method corresponds to writing down how to calculate it.

hfrick · 2025-08-29T10:44:58Z

learn/develop/filtro/index.qmd

+
+-   `results`: A slot for the results once the method is fitted. Initially, this is an empty data frame.
+
+For details on its constructor and its remaining properties, please refer to the package documentation.


It'b be nice to have a link here directly to the relevant docs.

hfrick · 2025-08-29T11:21:57Z

learn/develop/filtro/index.qmd

+  case_weights = NULL,
+  ...
+) {
+  # TODO Finish the rest of the function


Is this meant as a TODO for you now or do you plan to leave it in for the reader?

hfrick · 2025-08-29T11:24:23Z

learn/develop/filtro/index.qmd

+
+## Documenting S7 methods 
+
+Documentation for S7 methods is still a work in progress, and it seems no one currently knows the right approach. Here’s how we tackle it: 


Maybe something along the lines of "Documentation for S7 methods is still a work in progress, but our current best approach is as follows"?

Yes, I quoted Max too directly. 😅

hfrick · 2025-08-29T11:32:10Z

learn/develop/filtro/index.qmd

+```{r}
+#| eval: false
+#' @export
+S7::method(fit, class_score_aov) <- function(
+  object,
+  formula,
+  data,
+  case_weights = NULL,
+  ...
+) {
+  # TODO Finish the rest of the function using lm() and anova()
+
+  object@results <- res
+  object
+}
+```
+


I know I asked for examples previously but I'm leaning towards removing this one here because it's just one line that's relevant here. I think we can capture that in the text above. What do you think?

I truncated it down further, but I think it may still good to include the skeleton code here? Point of view from someone who is still navigating roxygen.

hfrick · 2025-08-29T11:32:49Z

learn/develop/filtro/index.qmd

+}
+```
+
+Instead, documentation is provided in the "Details" section and the "Estimating the scores" subsection of the documentation for the `score_aov_pval` object.  


The example for this one I find super helpful!

hfrick · 2025-08-29T11:35:21Z

learn/develop/filtro/index.qmd

+```{r}
+#| label: "si"
+#| echo: false
+small_session(pkgs)


For the final rendering, could you update filtro to the latest CRAN version? We have a convention of using only CRAN versions for these articles :)

franceslinyc · 2025-08-29T16:34:37Z

Still working!

franceslinyc · 2025-08-29T21:24:45Z

Thanks so much for the feedback @hfrick, I hope I didn't miss anything. @topepo or @EmilHvitfeldt, would you like to take a final look?

Initial commit

00faa50

This was referenced Aug 12, 2025

doc: Create score class object tidymodels/filtro#136

Open

Modify documentation along with tidymodel developer post tidymodels/filtro#163

Merged

franceslinyc added 2 commits August 18, 2025 10:49

Half way to first draft

8b92134

Hitting an error; I think it is related to PR 162 in filtro

99420d3

franceslinyc mentioned this pull request Aug 18, 2025

Use generics::fit() tidymodels/filtro#162

Merged

franceslinyc added 2 commits August 18, 2025 14:51

End of first draft

93e27c7

Checking with Max to see if we'd include discussion about fit()

8c42bae

franceslinyc marked this pull request as ready for review August 18, 2025 23:06

franceslinyc added 3 commits August 18, 2025 16:25

Minor editing

e304645

Add a subsection to discuss fit(); WIP

5cc29e5

WIP

dc992f2

franceslinyc added 4 commits August 19, 2025 09:09

Finalize a subsection to discuss fit()

f875e38

Finalize; Switch to filtro PR 162

f405fbc

Editing

5756248

Still can't run fit() : /

4c63d9f

Remove sorts

8ff9ae6

EmilHvitfeldt reviewed Aug 20, 2025

View reviewed changes

learn/develop/filtro/index.qmd Show resolved Hide resolved

franceslinyc added 2 commits August 20, 2025 09:36

Finalizing

9809084

Minor editing

5b48d78

franceslinyc requested a review from hfrick August 20, 2025 16:51

hfrick reviewed Aug 21, 2025

View reviewed changes

franceslinyc added 2 commits August 28, 2025 12:01

Revising based on Hannah's feedback; WIP

549822b

Finalizing

e6e3388

Elaborate on ANOVA F-test filter

e05fc22

hfrick reviewed Aug 29, 2025

View reviewed changes

Finalize

ffa95a2

franceslinyc added 2 commits August 29, 2025 13:41

Finalizing

ace06ad

READY!

a652720

franceslinyc requested a review from topepo August 29, 2025 21:25


		`r article_req_pkgs(pkgs)`

		You can construct new scoring objects using `class_score()`. This article is a guide to creating new scoring objects.


		## Scoring object

		All subclasses specific to the scoring method have a parent class named `class_score`. There are a few properties (attributes) for this object:


		## Scoring object specific to the scoring method

		We demonstrate how to create a custom scoring object specific to the given scoring method.


		We demonstrate how to create a custom scoring object specific to the given scoring method.

		As an example, let’s consider the ANOVA F-test filter.


		Documentation for S7 methods is still a work in progress, and it seems no one currently knows the right approach. Here’s how we tackle it:

		We re-export the `fit()` generic from another package. Instead of documenting each individual `fit()` method, we provide the details in the "Details" section and the "Estimating the scores" subsection of the documentation for the corresponding object.


		filtro is tidy tools to apply filter-based supervised feature selection methods. It provides functions to rank and select a specified proportion or a fixed number of features using built-in methods and the desirability function.

		Currently, there are 6 filters in filtro and many existing score objects. A list of existing scoring objects [can be found here](https://filtro.tidymodels.org/articles/filtro.html#available-score-objects-and-filter-methods). However, you might need to define your own scoring objects. This article serves as a guide to creating new scoring objects and computing feature scores before performing ranking and selection.


		The general procedure is to:

		1. Create a parent scoring object `class_score`, specifying fixed properties that are shared across all custom scoring objects.


		1. Create a parent scoring object `class_score`, specifying fixed properties that are shared across all custom scoring objects.

		2. Construct a custom scoring object `class_score_*`, adding additional, implementation-specific properties.


		2. Construct a custom scoring object `class_score_*`, adding additional, implementation-specific properties.

		3. Define the scoring method in `fit()`, which computes feature score. `fit()` refers to the custom scoring object from step 2 to use the appropriate method.


		- `results`: A slot for the results once the method is fitted. Initially, this is an empty data frame.

		For details on its constructor and its remaining properties, please refer to the package documentation.


		## Documenting S7 methods

		Documentation for S7 methods is still a work in progress, and it seems no one currently knows the right approach. Here’s how we tackle it:

Add a "Create your own score class object" page for filtro #117

Are you sure you want to change the base?

Add a "Create your own score class object" page for filtro #117

Conversation

franceslinyc commented Aug 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

franceslinyc commented Aug 18, 2025

Uh oh!

franceslinyc commented Aug 19, 2025

Uh oh!

franceslinyc commented Aug 19, 2025

Uh oh!

franceslinyc commented Aug 19, 2025

Uh oh!

Uh oh!

EmilHvitfeldt commented Aug 20, 2025

Uh oh!

hfrick left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

franceslinyc commented Aug 21, 2025

Uh oh!

hfrick left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

franceslinyc commented Aug 29, 2025

Uh oh!

franceslinyc commented Aug 29, 2025

Uh oh!

Uh oh!

franceslinyc commented Aug 12, 2025 •

edited

Loading