support for keras model returning props for one class

[nataliawojak]

Running lux on keras model which return probability for only one class an error occurs.

Code:
`#create daatframe with columns names as strings (LUX accepts only DataFrames withj string columns names)
iris = datasets.load_iris()
features = ['sepal_length','sepal_width','petal_length','petal_width']
target = 'class'
df_iris = pd.DataFrame(iris.data,columns=features)
df_iris[target] = iris.target
nn_model.fit(train[features],train[target])

#train classifier
train, test = train_test_split(df_iris)

input_shape=(train[features].shape[1],)
nn_model = Sequential()
nn_model.add(Dense(10, activation='relu',
input_shape=input_shape))
nn_model.add(Dense(10, activation='relu'))
nn_model.add(Dense(2, activation="sigmoid"))
nn_model.add(Dense(10, activation='relu'))
nn_model.add(Dense(1, activation="sigmoid"))
nn_model.compile(optimizer=Adam(0.01), loss='categorical_crossentropy')

#pick some instance from datasetr
iris_instance = train[features].sample(1).values
iris_instance

lux = LUX(predict_proba = nn_model.predict, neighborhood_size=20, max_depth=2, node_size_limit = 1, grow_confidence_threshold = 0 )
lux.fit(train[features], train[target], instance_to_explain=iris_instance,class_names=[0,1,2])

#see the justification of the instance being classified for a given class
lux.justify(np.array(iris_instance)) `

ERROR:

` InvalidParameterError Traceback (most recent call last)
in <cell line: 28>()
26
27 lux = LUX(predict_proba = nn_model.predict, neighborhood_size=20, max_depth=2, node_size_limit = 1, grow_confidence_threshold = 0 )
---> 28 lux.fit(train[features], train[target], instance_to_explain=iris_instance,class_names=[0,1,2])
29
30 #see the justification of the instance being classified for a given class

5 frames
/usr/local/lib/python3.10/dist-packages/sklearn/utils/_param_validation.py in validate_parameter_constraints(parameter_constraints, params, caller_name)
95 )
96
---> 97 raise InvalidParameterError(
98 f"The {param_name!r} parameter of {caller_name} must be"
99 f" {constraints_str}. Got {param_val!r} instead."

InvalidParameterError: The 'n_neighbors' parameter of NearestNeighbors must be an int in the range [1, inf) or None. Got 0 instead. `