-
-
Notifications
You must be signed in to change notification settings - Fork 26
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New functions for investigating predictor rankings #406
Conversation
Add new functions `ranking()` and `props()`, making output element `pct_solution_terms_cv` of `vsel` objects obsolete (but requiring the addition of output element `solution_terms_cv` to `vsel` objects). Also add new `plot()` methods for visualizing the output of `props()`.
the non-sampled observations consist of `NA`s.
`cv_varsel()`, speed-up the `cv_varsel()` example, and add an example for `plot.props()` (thereby also implying examples for `ranking()` and `props()`).
instead of throwing an informational message and returning `NULL`. This avoids that users call `plot()` on `NULL`, which throws the opaque error `need finite 'xlim' values`.
fold-wise predictor rankings, an error is thrown instead of `NULL` being returned.
Why Otherwise, the illustration looks good |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I checked the documentation parts, too
|
We now agreed to rename |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As discussed, ok now
This PR starts to address #289.
In particular, this adds the new functions
ranking()
andprops()
, making output elementpct_solution_terms_cv
ofvsel
objects obsolete (but requiring the addition of output elementsolution_terms_cv
tovsel
objects). This also adds a newplot()
method for visualizing the output ofprops()
(more precisely, twoplot()
methods are added, butplot.vsel()
is only a shortcut for first applyingranking.vsel()
and thenplot.ranking()
). For the user-facing changes, see the changes inNEWS.md
.Quick illustration of the new functions:
Dummy ranking proportions for illustrating the color gradient:
As mentioned above, this PR is only a first step towards resolving #289 completely. In the future, I will:
print.vselsummary()
table, containing the diagonal of the matrix of the non-cumulated ranking proportions.plot.vsel()
predictive performance plot, again with the diagonal of the matrix of the non-cumulated ranking proportions. This will be done with the option to restrict this solution path and the ranking proportions to the first few terms. By default, I think we should not restrict them, for consistency with theprint.vselsummary()
table.