| Type: | Package |
| Title: | Perform Inference on Summaries of Longitudinal Algorithm-Agnostic Variable Importance |
| Version: | 1.0.0 |
| Description: | Calculate point estimates of and valid confidence intervals for longitudinal summaries of nonparametric, algorithm-agnostic variable importance measures. For more details, see Williamson et al. (2024) <doi:10.48550/arXiv.2311.01638>. |
| License: | MIT + file LICENSE |
| Imports: | vimp |
| Suggests: | knitr, rmarkdown, testthat, SuperLearner |
| URL: | https://bdwilliamson.github.io/lvimp/ |
| BugReports: | https://github.com/bdwilliamson/lvimp/issues |
| VignetteBuilder: | knitr |
| RoxygenNote: | 7.3.2 |
| Encoding: | UTF-8 |
| NeedsCompilation: | no |
| Packaged: | 2025-12-01 17:05:16 UTC; L107067 |
| Author: | Brian D. Williamson
 [aut, cre] |
| Maintainer: | Brian D. Williamson <brian.d.williamson@kp.org> |
| Repository: | CRAN |
| Date/Publication: | 2025-12-06 12:20:02 UTC |
lvimp: Perform Inference on Summaries of Longidutinal Algorithm-Agnostic Variable Importance
Description
Calculate point estimates of and valid confidence intervals for longitudinal summaries of nonparametric, algorithm-agnostic variable importance measures. For more details, see Williamson et al. (2024) doi:10.48550/arXiv.2311.01638.
Authors
Maintainer: Brian Williamson https://bdwilliamson.github.io/
Methodology authors:
Brian D. Williamson
Susan M. Shortreed
See Also
Manuscripts: (to appear)
Imports
The packages that we import either make internal code nice (dplyr, magrittr, tibble, rlang, data.table), or are used for estimating and performing inference on cross-sectional variable importance (vimp).
We suggest several other packages: ggplot2 and cowplot help with plotting variable importance estimates; testthat and covr help with unit tests; and knitr and rmarkdown help with the vignettes and examples.
Author(s)
Maintainer: Brian D. Williamson brian.d.williamson@kp.org (ORCID)
See Also
Useful links:
Format a lvim object
Description
Format a lvim object
Usage
## S3 method for class 'lvim'
format(x, digits = 3, ...)
Arguments
x |
the |
digits |
the number of digits to format to |
... |
other options, see the generic |
Value
A formatted lvim object for printing.
Create a Longitudinal Variable Importance Object
Description
Create a longitudinal variable importance object from several constituent cross-sectional variable importance objects.
Usage
lvim(vim_list = list(), timepoints = numeric())
Arguments
vim_list |
a list of individual, cross-sectional variable importance objects. Assumed to be in order over time. |
timepoints |
a numeric vector of timepoints of interest |
Value
an object of class lvim
Area Under the Variable Importance Trajectory
Description
Compute a nonparametric estimate of (and efficient influence function for) the area under the longitudinal variable importance trajectory (AUTC) over a contiguous subset of the time series.
Usage
lvim_autc(
lvim,
indices = 1:length(lvim),
interpolator = "linear",
delta = 0,
...
)
Arguments
lvim |
an object of class |
indices |
a numeric vector indicating the contiguous subset of the time series |
interpolator |
a string indicating the type of interpolator used to take the area under the trajectory |
delta |
null hypothesis value |
... |
other arguments to be passed to the interpolator function |
Value
The lvim object, with point estimates, CIs, and p-values
related to the area under the trend in variable importance filled in.
Average Longitudinal Variable Importance
Description
Compute a nonparametric estimate of (and efficient influence function for) the average longitudinal variable importance over a contiguous subset of the time series.
Usage
lvim_average(lvim, indices = 1:length(lvim), delta = 0)
Arguments
lvim |
an object of class |
indices |
a numeric vector indicating the contiguous subset of the time series |
delta |
null hypothesis value |
Value
The lvim object, with point estimates, CIs, and p-values
related to the average variable importance filled in.
Linear Trend in the Longitudinal Variable Importance Trajectory
Description
Compute a nonparametric estimate of (and efficient influence function for) the linear trend in the longitudinal variable importance over a contiguous subset of the time series.
Usage
lvim_trend(lvim, indices = 1:length(lvim), delta = 0)
Arguments
lvim |
an object of class |
indices |
a numeric vector indicating the contiguous subset of the time series |
delta |
null hypothesis value |
Value
The lvim object, with point estimates, CIs, and p-values
related to the linear trend in variable importance filled in.
Print a lvim object
Description
Print a lvim object
Usage
## S3 method for class 'lvim'
print(x, ...)
Arguments
x |
the |
... |
other options, see the generic |
Value
No return value, called for side effects.