feat!: aif360 integration

requirements.txt

@@ -1,2 +1,2 @@
-scikit-learn==1.4.1.post1
-numpy==1.26.4
+aif360==0.6.0
+pandas==2.2.1

test/__init__.py

@@ -1,52 +1,10 @@
 import numpy as np
 import numpy.random as random
-import typing
+import pandas as pd
 
 
-SEED = 1
-random.seed(SEED)
-
-
-def uniform_binary_dataset(rows: int = 1000, columns: int = 2) -> np.array:
-    return random.uniform(0, 1, size=(rows, columns)).round().astype(int)
-
-
-def uniform_binary_dataset_gt(rows: int = 1000, columns: int = 2) -> np.array:
-    xs = uniform_binary_dataset(rows, 1)
-    labels = uniform_binary_dataset(rows, 1) 
-    noise = random.choice([0, 1], p=[0.8, 0.2], size=(rows, 1))
-    preds = abs(labels-noise)
-    if columns > 2:
-        data = []
-        for _ in range(columns - 2):
-            data.append(uniform_binary_dataset(rows, 1))
-        data = np.concatenate(data, axis=1)
-        return np.concatenate((data, xs, labels, preds), axis=1) 
-    else:
-        return np.concatenate((xs, labels, preds), axis=1)
-
-def skewed_binary_dataset_gt(rows: int = 1000, columns: int = 2, p: float = 0.8) -> np.array:
-    xs = uniform_binary_dataset(rows, 1)
-    preds = np.array([bernoulli(p)[0] * x for x in xs])
-    labels = uniform_binary_dataset(rows, 1)
-    if columns > 2:
-        data = []
-        for _ in range(columns - 2):
-            data.append(uniform_binary_dataset(rows, 1))
-        data = np.concatenate(data, axis=1)
-        return np.concatenate((data, xs, labels, preds), axis=1) 
-    else:
-        return np.concatenate((xs, labels, preds), axis=1)
-
-def bernoulli(p: float, size: typing.Tuple[int, int] = (1,)) -> np.array:
-    assert 0 <= p <= 1, "p must be in [0, 1]"
-    return (random.uniform(0, 1, size=size) < p).astype(int)
-
-
-def skewed_binary_dataset(rows: int = 1000, columns: int = 2, p: float = 0.8) -> np.array:
-    xs = uniform_binary_dataset(rows, 1)
-    cols = []
-    for _ in range(columns - 1):
-        ys = np.array([bernoulli(p)[0] * x for x in xs])
-        cols.append(ys)
-    return np.concatenate([xs] + cols, axis=1)
+def binary_label_dataset(rows: int = 1000) -> pd.DataFrame:
+    features = random.uniform(0, 1, size=(rows, 1))
+    labels = random.uniform(0, 1, size=(rows, 1)).astype(int)
+    data = np.concatenate([features] + [labels], axis=1)
+    return pd.DataFrame(data, columns=['feat', 'label'])