load_sample_from_dataset(y_desired_index, dataset)
# Training a ML model
for algorithm in algorithms:
if verbose:
print(f'Training using {algorithm}.')
model = train_ml_model(X, y, algorithm)
if verbose:
print(f'Starting counterfactual generation.')
pareto_front, pareto_set, pareto_algorithms = \
generate_counterfactuals_classification_simple(model, X_current, \
y_desired, immutable_column_indexes, upper_bounds, \
lower_bounds, categorical_columns, integer_columns, \
n_gen=100, pop_size=100, max_changed_vars=10, \
verbose=verbose, seed=0)
trained_models.append([
dataset, algorithm, pareto_front, pareto_set, pareto_algorithms
])
return trained_models
results_proba = train_classification_proba(verbose=True)
results_simple = train_classification_simple(verbose=True)
view_results(results_proba)
view_results(results_simple)
upper_bounds = np.max(X, axis=0)
lower_bounds = np.min(X, axis=0)
# Training a ML model
for algorithm in algorithms:
if verbose:
print(f'Training using {algorithm}')
model = train_ml_model(X, y, algorithm)
if verbose:
print(f'Starting counterfactual generation.')
pareto_front, pareto_set, pareto_algorithms = \
generate_counterfactuals_regression(model, \
scaler.transform(X_current.reshape(1, -1)), \
y_desired, immutable_column_indexes, \
y_acceptable_range, upper_bounds, lower_bounds, \
categorical_columns, integer_columns, n_gen=50, \
pop_size=300, max_changed_vars=10, verbose=verbose, seed=0)
trained_models.append([
dataset, algorithm, pareto_front, pareto_set, pareto_algorithms
])
return np.array(trained_models)
results = train_regression(verbose=True)
view_results(results)
upper_bounds, lower_bounds, y_acceptable_range, \
categorical_columns, integer_columns = \
load_sample_from_dataset(y_desired_index, dataset)
# Training a ML model
for algorithm in algorithms:
if verbose:
print(f'Training using {algorithm}.')
model = train_ml_model(X, y, algorithm)
if verbose:
print(f'Starting counterfactual generation.')
pareto_front, pareto_set, pareto_algorithms = \
generate_counterfactuals_classification_proba(model, \
X_current, y_desired, immutable_column_indexes, \
y_acceptable_range, upper_bounds, lower_bounds, \
categorical_columns, integer_columns, n_gen=50, \
pop_size=100, max_changed_vars=5, verbose=verbose, seed=0)
trained_models.append([
dataset, algorithm, pareto_front, pareto_set, pareto_algorithms
])
return np.array(trained_models)
results_proba = train_classification_proba(verbose=True)
view_results(results_proba)