Package: limestab 0.1.0

Vilda Purutcuoglu

limestab: Stable Local Interpretable Model-Agnostic Explanations (LIME, OptiLIME, BayLIME, ALIME)

Provides four local explanation methods for tabular data: LIME, OptiLIME, BayLIME, and ALIME, together with the stability metrics used to compare them across repeated explanations of the same instance: the Coefficient Stability Index (CSI), the Variable Stability Index (VSI), Median Rank Matching Stability (MRMS), Point-Based Stability Index (PCSI), and Interval Coefficient Stability Index (ICSI). The methodology accompanies Toparlak and Purutcuoglu, "A Median-Based Framework for Assessing the Stability in XAI Tools" (manuscript under revision); the citation will be updated upon publication.

Authors:Simay Toparlak [aut], Vilda Purutcuoglu [aut, cre]

limestab_0.1.0.tar.gz
limestab_0.1.0.zip(r-4.7-any)limestab_0.1.0.zip(r-4.6-any)
limestab_0.1.0.tar.gz(r-4.7-any)limestab_0.1.0.tar.gz(r-4.6-any)
limestab_0.1.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION
card.svg |card.png
limestab/json (API)

# Install 'limestab' in R:
install.packages('limestab', repos = c('https://cran.r-universe.dev', 'https://cloud.r-project.org'))

On CRAN:

Conda:

This package does not link to any Github/Gitlab/R-forge repository. No issue tracker or development information is available.

1.70 score 28 exports 0 dependencies

Last updated from:bafba625b4. Checks:6 OK. Indexed: yes.

TargetResultTimeFilesSyslog
linux-develOK142
source / vignettesOK188
linux-releaseOK103
windows-develOK82
windows-releaseOK92
wasm-releaseOK93

Exports:alime_explain_instancealime_explainerbayesian_optimisationbayesian_ridge_fitbaylime_explain_instancebaylime_explainercalculate_distancescompute_cov_metricscompute_csicompute_icsicompute_mad_stdevscompute_mrmscompute_pcsicompute_vsicompute_wls_stdevsdefault_kernel_widthfit_gpfit_weighted_ridgegenerate_perturbationslime_check_stabilitylime_confidence_intervalslime_explain_instancelime_explainerlime_kerneloptilimeoptilime_scoreselect_featuresweighted_r2

Dependencies:

Readme and manuals

Help Manual

Help pageTopics
limestab: Stable LIME, OptiLIME, BayLIME and ALIMElimestab-package limestab
Explain a single instance with ALIMEalime_explain_instance
Create an ALIME explaineralime_explainer
Bayesian optimisation of a scalar objectivebayesian_optimisation
Bayesian ridge regression (Tipping 2001 / MacKay 1992)bayesian_ridge_fit
Explain a single instance with BayLIMEbaylime_explain_instance
Create a BayLIME explainerbaylime_explainer
Euclidean distance of every row to an instancecalculate_distances
Coefficient-of-variation metrics across repeated runscompute_cov_metrics
Coefficient Stability Index (CSI)compute_csi
Interval-Based Coefficient Stability Index (ICSI)compute_icsi
Robust standard deviation from repeated coefficientscompute_mad_stdevs
Median Rank Matching Stability of features across repeated runscompute_mrms
PCSI stability indicescompute_pcsi
Variable Stability Index (VSI)compute_vsi
Standard deviations of weighted ridge coefficientscompute_wls_stdevs
Default LIME kernel width for tabular datadefault_kernel_width
Fit a 1-D Gaussian process (Matern 3/2 kernel)fit_gp
Weighted ridge regression with an unpenalised interceptfit_weighted_ridge
Generate a perturbed neighbourhood in standardised feature spacegenerate_perturbations
Repeat an explanation and compute every stability metriclime_check_stability
Confidence intervals for a LIME-style explanationlime_confidence_intervals
Explain a single instance with LIMElime_explain_instance
Create a LIME explainerlime_explainer
LIME exponential kernellime_kernel
OptiLIME: choose the LIME kernel width by Bayesian optimisationoptilime
OptiLIME objective (reflected local R-squared)optilime_score
Predict method for optilime_gp objectspredict.optilime_gp
Select the features used by a LIME-style local modelselect_features
Weighted R-squaredweighted_r2