BayLIME extension (visualime.baylime)
BayLIME (Bayesian Local Interpretable Model-Agnostic Explanations) [1] is an extension of LIME [2] that exploits prior knowledge and Bayesian reasoning.
[1] Zhao et al. (2021): BayLIME: Bayesian Local Interpretable Model-Agnostic Explanations. arXiv:2012.03058
[2] Ribeiro et al. (2016): “Why Should I Trust You?”: Explaining the Predictions of Any Classifier. arXiv:1602.04938
- class visualime.baylime.BayesianRidgeFixedLambda(*, epsilon: float = 1e-06, large_number: float = 1000000000.0, **kwargs)[source]
Bases:
BayesianRidge
BayesianRidge model with fixed parameter lambda. This is equivalent to the “partial informative priors” option in the BayLIME framework.
The value for lambda_init will be treated as a constant. The parameter epsilon specifies the maximum amount of change allowed for lambda. A warning will be issued if this amount is exceeded. To reduce the amount of change in lambda, increase large_number.
See the documentation for
sklearn.linear_model.BayesianRidge()
for a list of available parameters.Methods
fit
(X, y[, sample_weight])Fit the model.
get_params
([deep])Get parameters for this estimator.
predict
(X[, return_std])Predict using the linear model.
score
(X, y[, sample_weight])Return the coefficient of determination of the prediction.
set_params
(**params)Set the parameters of this estimator.
- fit(X, y, sample_weight=None)[source]
Fit the model.
- Parameters:
- X
ndarray
ofshape
(n_samples
,n_features
) Training data.
- y
ndarray
ofshape
(n_samples,) Target values. Will be cast to X’s dtype if necessary.
- sample_weight
ndarray
ofshape
(n_samples,), default=None Individual weights for each sample.
New in version 0.20: parameter sample_weight support to BayesianRidge.
- X
- Returns:
- self
object
Returns the instance itself.
- self
- class visualime.baylime.BayesianRidgeFixedAlphaLambda(*, epsilon: float = 1e-06, large_number: float = 1000000000.0, **kwargs)[source]
Bases:
BayesianRidge
BayesianRidge model with fixed parameters alpha and lambda. This is equivalent to the “full informative priors” option in the BayLIME framework.
The values for alpha_init and lambda_init will be treated as constants.
The parameter epsilon specifies the maximum amount of change allowed for alpha and lambda. A warning will be issued if this amount is exceeded. To reduce the amount of change, increase large_number.
See the documentation for
sklearn.linear_model.BayesianRidge()
for a list of available parameters.Methods
fit
(X, y[, sample_weight])Fit the model.
get_params
([deep])Get parameters for this estimator.
predict
(X[, return_std])Predict using the linear model.
score
(X, y[, sample_weight])Return the coefficient of determination of the prediction.
set_params
(**params)Set the parameters of this estimator.
- fit(X, y, sample_weight=None)[source]
Fit the model.
- Parameters:
- X
ndarray
ofshape
(n_samples
,n_features
) Training data.
- y
ndarray
ofshape
(n_samples,) Target values. Will be cast to X’s dtype if necessary.
- sample_weight
ndarray
ofshape
(n_samples,), default=None Individual weights for each sample.
New in version 0.20: parameter sample_weight support to BayesianRidge.
- X
- Returns:
- self
object
Returns the instance itself.
- self