Classifiers utilities¶
-
clf_utilities.create_clf_params_product_generator(params_grid)[source]¶ Generates all possible combinations of classifier’s hyperparameters values.
- Parameters
params_grid (dict) – Contains classifier’s hyperparameters names as keys and the correspoding search space as values
- Yields
dict – Contains a classifier’s hyperparameters configuration
-
clf_utilities.create_feature_sets_generator(fold_path)[source]¶ Creates a generator yielding features sets names.
- Parameters
fold_path (str) – Path to read features sets
- Yields
list – pairs of X_train, X_test features sets names
-
clf_utilities.evaluate(y_test, y_pred)[source]¶ Evaluates model predictions through a series of metrics.
- Parameters
y_test (numpy.ndarray) – True labels
y_pred (numpy.ndarray) – Predicted labels
- Returns
Contains metrics names as keys and the corresponding values as values
- Return type
dict
-
clf_utilities.get_top_k_predictions(model, X_test)[source]¶ Makes predictions utilizing model over X_test.
- Parameters
model (object) – The model to be used for predictions
X_test (numpy.ndarray) – The test features array
- Returns
Contains predictions in (label, score) pairs
- Return type
list
-
clf_utilities.inverse_transform_labels(encoder, k_preds)[source]¶ Utilizes encoder to transform encoded labels back to the original strings.
- Parameters
encoder (sklearn.preprocessing.LabelEncoder) – The encoder to be utilized
k_preds (list) – Contains predictions in (label, score) pairs
- Returns
Contains predictions in (label, score) pairs, where label is now in the original string format
- Return type
list
-
clf_utilities.is_valid(clf_name)[source]¶ Checks whether clf_name is a valid classifier’s name with respect to the experiment setup.
- Parameters
clf_name (str) – Classifier’s name
- Returns
Returns True if given classifier’s name is valid
- Return type
bool
-
clf_utilities.k_accuracy_score(y_test, k_best)[source]¶ Measures the defined k-accuracy metric. For each poi, a successful prediction is considered if true label appears in the top k labels predicted by the model,
- Parameters
y_test (numpy.ndarray) – True labels
k_best (numpy.ndarray) – Top k predicted labels
- Returns
The k accuracy score
- Return type
float
-
clf_utilities.normalize_scores(scores)[source]¶ Normalizes predictions scores to a probabilities-like format.
- Parameters
scores (list) – Contains the predictions scores as predicted by the model
- Returns
The normalized scores
- Return type
list
-
clf_utilities.train_classifier(clf_name, X_train, y_train)[source]¶ Trains a classifier through grid search.
- Parameters
clf_name (str) – Classifier’s name to be trained
X_train (numpy.ndarray) – Train features array
y_train (numpy.ndarray) – Train labels array
- Returns
The trained classifier
- Return type
object