"feature_weights" not working

[DarthZonke]

I cannot use "feature_weights" on DICE, I am getting the error below, please assist:

Code:
query_instance = denied_loans.drop(columns="predicted")
dice_exp = exp.generate_counterfactuals(query_instance,
total_CFs=1,
desired_class="opposite",
features_to_vary=vary_feature,
proximity_weight=1,
diversity_weight=0,
categorical_penalty=1, .
feature_weights=feature_weights,
random_seed=42,
verbose=True,
method="genetic")

dice_exp.visualize_as_dataframe()

Error:

---

TypeError Traceback (most recent call last)
in <cell line: 2>()
1 query_instance = denied_loans.drop(columns="predicted")
----> 2 dice_exp = exp.generate_counterfactuals(query_instance,
3 total_CFs=1,
4 desired_class="opposite",
5 features_to_vary=vary_feature,

/usr/local/lib/python3.10/dist-packages/dice_ml/explainer_interfaces/explainer_base.py in generate_counterfactuals(self, query_instances, total_CFs, desired_class, desired_range, permitted_range, features_to_vary, stopping_threshold, posthoc_sparsity_param, proximity_weight, sparsity_weight, diversity_weight, categorical_penalty, posthoc_sparsity_algorithm, verbose, **kwargs)
184 for query_instance in tqdm(query_instances_list):
185 self.data_interface.set_continuous_feature_indexes(query_instance)
--> 186 res = self._generate_counterfactuals(
187 query_instance, total_CFs,
188 desired_class=desired_class,

TypeError: DiceRandom._generate_counterfactuals() got an unexpected keyword argument 'feature_weights

[mpereda]

Hi!

I have experienced the same problem. The source code regarding the method for generating CFs shows that the argument 'feature_weights' is not implemented. It just uses the default option which is the MADs https://interpret.ml/DiCE/_modules/dice_ml/explainer_interfaces/dice_random.html

However, user-defined feature_weights can used with the genetic algorithm method. Following the example in https://interpret.ml/DiCE/notebooks/DiCE_with_advanced_options.html, you can have:

exp2= dice_ml.Dice(d, m, method="genetic")
dice_exp2 = exp2.generate_counterfactuals(query_df, total_CFs=4, desired_class="opposite", feature_weights=feature_weights)
dice_exp2.visualize_as_dataframe(show_only_changes=True)

and the output will be:
age workclass education marital_status occupation race gender hours_per_week income
0 - - Doctorate Married White-Collar - Male - 1
0 - - Bachelors Married Blue-Collar - Male - 1
0 - - Some-college Married White-Collar - Male - 1
0 26 - Bachelors Married White-Collar - - - 1

I hope this helps!
Best,
María