src.ml_cfexplainer.explainer package

Submodules

src.ml_cfexplainer.explainer.cf_adult module

src.ml_cfexplainer.explainer.cf_adult.device = device(type='cpu')

30 Number of objective function: 2

Type

Number of variable

class src.ml_cfexplainer.explainer.cf_adult.CF_Adult(x0, d, features, pred_model, dfencoder_model, pos_proto, x6, x7, n_var=8, **kwargs)[source]

Bases: pymoo.model.problem.Problem

src.ml_cfexplainer.explainer.cf_credit module

src.ml_cfexplainer.explainer.cf_sangiovese module

src.ml_cfexplainer.explainer.cf_simplebn module

class src.ml_cfexplainer.explainer.cf_simplebn.CF_SimpleBN(x0, pred_model, dfencoder_model, scm_model, proto, col, con_index=None, cat_index=None, n_var=3, **kwargs)[source]

Bases: pymoo.model.problem.Problem

src.ml_cfexplainer.explainer.distance module

class src.ml_cfexplainer.explainer.distance.Distance(x0, xcf, pred_model, dfencoder_model, cat_index=None, con_index=None, dict_cat_index=None)[source]

Bases: object

Distance class

pure_distance()[source]
Parameters

  • x0 (int) – original instance

  • xcf (int) – counterfactual instance

Returns

distance between two instance

Return type

float

continous_dist()[source]
Parameters

  • x0 (int) – original instance

  • xcf (int) – counterfactual instance

Returns

distance between two instance

Return type

float

logloss()[source]
latent_distance(z0, zcf)[source]
Parameters

  • x0 (int) – original instance

  • xcf (int) – counterfactual instance

Returns

distance between two instance

Return type

float

two_cate_dist(model, k, cat1, cat2)[source]
Parameters

  • x0 (int) – original instance

  • xcf (int) – counterfactual instance

Returns

distance between two instance

Return type

float

cat_representation_dist()[source]

Compute the categorical distance in latent space

Returns

DESCRIPTION

Return type

TYPE

proto_loss(zcf, proto)[source]
Parameters

  • zcf (TYPE) – DESCRIPTION

  • proto (TYPE) – DESCRIPTION

Returns

DESCRIPTION

Return type

TYPE

constraints_loss()[source]

Compute the constraint loss: Age have to be larger than original instance

Returns

DESCRIPTION

Return type

TYPE

compute_yloss(ycf, prediction_model, d)[source]

Compute the prediction loss function

Parameters

  • xcf (TYPE) – DESCRIPTION

  • ycf (TYPE) – DESCRIPTION

  • prediction_model (TYPE) – DESCRIPTION

  • d (TYPE) – DESCRIPTION

Returns

DESCRIPTION

Return type

TYPE

causal_loss_adult()[source]
Returns

DESCRIPTION

Return type

TYPE

causal_loss_sangio(xcf, x0)[source]
social_cost()[source]
cross_entropy(targets=1, epsilon=1e-12)[source]

Computes cross entropy between targets (encoded as one-hot vectors) and predictions. Input: predictions (N, k) ndarray

targets (N, k) ndarray

Returns: scalar

src.ml_cfexplainer.explainer.prototype module

Created on Thu Dec 3 11:06:31 2020

@author: trduong

src.ml_cfexplainer.explainer.prototype.find_proto(original_latent, pos_latent, neg_latent, k_instance)[source]
src.ml_cfexplainer.explainer.prototype.get_pos_neg_latent(prediction, z_representation)[source]

Module contents