Data and reproducibility¶
Input dictionary¶
| Input | Interpretation |
|---|---|
X |
Numeric two-dimensional array: observations in rows, predictors in columns |
y |
Numeric response vector matching the rows of X |
New X |
Same predictor columns and order as training |
seed |
Fold-construction seed in the relevant CV helper/estimator |
The estimator source implements its documented internal transformations. This presentation adds no preprocessing or automatic data cleaning.
Data sources and access¶
The repository contains no application datasets. The study cites these original providers; using the software does not transfer ownership of their data.
| Application data | Original provider |
|---|---|
| SLID | R carData package |
| NOx | R robustbase package |
| QSAR fish toxicity | UCI Machine Learning Repository |
| Concrete compressive strength | UCI Machine Learning Repository |
The paper reports the analysis protocols. Application workflows and study materials can be requested from the authors for research verification. This site does not claim that installing the package reproduces all paper results.
Reproducible installation check¶
From a reviewed checkout and a Python 3.13 environment with the package installed:
The existing test uses a small synthetic sample and checks finite predictions for the two principal classes. Record the source revision, Python version, dependency versions, configuration and seeds with any analysis. Neither this test nor a successful site build validates a production simulation study.
Documentation environment¶
Documentation uses MkDocs/Material and a separate requirements-docs.txt.
It does not import the estimator or execute fitting during site generation.
python3.13 -m venv .venv-docs
. .venv-docs/bin/activate
python -m pip install -r requirements-docs.txt
mkdocs build --strict
The output is static HTML. Publication is a separate maintainer action.