merge conflicts and broken tests

cleanlab/count.py

@@ -1359,7 +1359,7 @@ class 0, 1, ..., K-1. `pred_probs` should have been computed using 3 (or
     confident_thresholds : np.ndarray
       An array of shape ``(K, )`` where K is the number of classes."""
 
-    labels = cleanlab.internal.validation.labels_to_array(labels)
+    labels = labels_to_array(labels)
     all_classes = range(pred_probs.shape[1])
     unique_classes = get_unique_classes(labels)
     BIG_VALUE = 2

cleanlab/filter.py

@@ -242,7 +242,6 @@ class 0, 1, ..., K-1.
     else:
         assert n_jobs >= 1
 
-    # Number of examples in each class of labels
     if multi_label:
         return _find_label_issues_multilabel(
             labels,
@@ -258,10 +257,7 @@ class 0, 1, ..., K-1.
             verbose,
         )
 
-    else:
-        label_counts = value_counts(labels)
-
-    # Number of classes
+    # Else this is standard multi-class classification
     K = get_num_classes(
         labels=labels, pred_probs=pred_probs, label_matrix=confident_joint, multi_label=multi_label
     )

cleanlab/internal/multilabel_utils.py

@@ -58,7 +58,7 @@ def get_onehot_num_classes(labels, pred_probs=None):
     """Returns OneHot encoding of MultiLabel Data, and number of classes"""
     num_classes = get_num_classes(labels=labels, pred_probs=pred_probs)
     try:
-        y_one = int2onehot(labels)
+        y_one = int2onehot(labels, K=num_classes)
     except TypeError:
         raise ValueError(
             "wrong format for labels, should be a list of list[indices], please check the documentation in find_label_issues for further information"

cleanlab/internal/util.py

@@ -264,6 +264,7 @@ def round_preserving_row_totals(confident_joint) -> np.ndarray:
     ).astype(int)
 
 
+# TODO: move to multilabel_utils.py
 def int2onehot(labels, K) -> np.ndarray:
     """Convert list of lists to a onehot matrix for multi-labels
 
@@ -282,6 +283,7 @@ def int2onehot(labels, K) -> np.ndarray:
     return mlb.fit_transform(labels)
 
 
+# TODO: move to multilabel_utils.py
 def onehot2int(onehot_matrix) -> list:
     """Convert a onehot matrix for multi-labels to a list of lists of ints
 

tests/test_filter_count.py

@@ -117,7 +117,8 @@ def make_multilabel_data(
     seed=1,
 ):
     np.random.seed(seed=seed)
-    m = len(means) + len(
+    num_classes = len(means)
+    m = num_classes + len(
         box_multilabels
     )  # number of classes by treating each multilabel as 1 unique label
     n = sum(sizes)
@@ -172,7 +173,7 @@ def make_multi(X, Y, bx1, by1, bx2, by2, label_list):
     ps = np.bincount(labels_idx) / float(len(labels_idx))
     inv = compute_inv_noise_matrix(py, noise_matrix, ps=ps)
 
-    y_train = int2onehot(noisy_labels)
+    y_train = int2onehot(noisy_labels, K=num_classes)
     clf = MultiOutputClassifier(LogisticRegression())
     pyi = cross_val_predict(clf, X_train, y_train, method="predict_proba")
     pred_probs = np.zeros(y_train.shape)
@@ -310,7 +311,7 @@ def test_calibrate_joint_multilabel():
         labels=dataset["labels"],
         multi_label=True,
     )
-    y_one = int2onehot(dataset["labels"])
+    y_one = int2onehot(dataset["labels"], K=dataset["pred_probs"].shape[1])
     # Check calibration
     for class_num in range(0, len(calibrated_cj)):
         label_counts = np.bincount(y_one[:, class_num])

tests/test_util.py

@@ -79,8 +79,9 @@ def test_round_preserving_sum():
 
 
 def test_one_hot():
+    num_classes = 4
     labels = [[0], [0, 1], [0, 1], [2], [0, 2, 3]]
-    assert onehot2int(int2onehot(labels)) == labels
+    assert onehot2int(int2onehot(labels, K=num_classes)) == labels
 
 
 def test_num_unique():