| |
| |
|
|
| from collections import defaultdict, namedtuple |
| import numpy as np |
| import pandas as pd |
| from fairlearn.postprocessing._threshold_operation import ThresholdOperation |
| from fairlearn.postprocessing._constants import SCORE_KEY, LABEL_KEY, SENSITIVE_FEATURE_KEY |
|
|
|
|
| sensitive_feature_names_ex1 = ["A", "B", "C"] |
| sensitive_features_ex1 = [x for x in 'AAAAAAA' 'BBBBBBB' 'CCCCCC'] |
| sensitive_feature_names_ex2 = ["x", "Y"] |
| sensitive_features_ex2 = [x for x in 'xxxYYYY' 'xYYYYYx' 'YYYYYY'] |
| labels_ex = [int(x) for x in '0110100' '0010111' '000111'] |
| degenerate_labels_ex = [int(x) for x in '0000000' '0000000' '000000'] |
| scores_ex = [int(x) for x in '0011233' '0001111' '011112'] |
|
|
| LabelAndPrediction = namedtuple('LabelAndPrediction', 'label prediction') |
|
|
|
|
| class ExamplePredictor(): |
| def predict(self, X): |
| return scores_ex |
|
|
|
|
| class ExampleNotPredictor(): |
| pass |
|
|
|
|
| class ExampleEstimator(): |
| def fit(self, X, Y): |
| pass |
|
|
| def predict(self, X): |
| return scores_ex |
|
|
|
|
| class ExampleNotEstimator1(): |
| def fit(self, X, Y): |
| pass |
|
|
|
|
| class ExampleNotEstimator2(): |
| def predict(self, X): |
| pass |
|
|
|
|
| def _get_grouped_data_and_base_points(sensitive_feature_value): |
| data = pd.DataFrame({ |
| SENSITIVE_FEATURE_KEY: sensitive_features_ex1, |
| SCORE_KEY: scores_ex, |
| LABEL_KEY: labels_ex}) |
| grouped_data = data.groupby(SENSITIVE_FEATURE_KEY).get_group(sensitive_feature_value) \ |
| .sort_values(by=SCORE_KEY, ascending=False) |
| x_grid = np.linspace(0, 1, 100) |
|
|
| if sensitive_feature_value == "A": |
| expected_roc_points = pd.DataFrame({ |
| "x": [0, 0.25, 0.5, 0.5, 1], |
| "y": [0, 1/3, 2/3, 1, 1], |
| "operation": [ThresholdOperation('>', np.inf), |
| ThresholdOperation('<', 0.5), |
| ThresholdOperation('<', 1.5), |
| ThresholdOperation('<', 2.5), |
| ThresholdOperation('>', -np.inf)] |
| }) |
| ignore_for_base_points = [1, 2] |
|
|
| if sensitive_feature_value == "B": |
| expected_roc_points = pd.DataFrame({ |
| "x": [0, 1/3, 1], |
| "y": [0, 3/4, 1], |
| "operation": [ThresholdOperation('>', np.inf), |
| ThresholdOperation('<', 0.5), |
| ThresholdOperation('>', -np.inf)] |
| }) |
| ignore_for_base_points = [] |
|
|
| if sensitive_feature_value == "C": |
| expected_roc_points = pd.DataFrame({ |
| "x": [0, 0, 2/3, 1], |
| "y": [0, 1/3, 1, 1], |
| "operation": [ThresholdOperation('>', np.inf), |
| ThresholdOperation('<', 0.5), |
| ThresholdOperation('<', 1.5), |
| ThresholdOperation('>', -np.inf)] |
| }) |
| ignore_for_base_points = [0] |
|
|
| return grouped_data, expected_roc_points, ignore_for_base_points, x_grid |
|
|
|
|
| def _get_predictions_by_sensitive_feature(adjusted_predictor, sensitive_features, scores, labels): |
| labels_and_predictions = defaultdict(list) |
| for i in range(len(sensitive_features)): |
| labels_and_predictions[sensitive_features[i]].append( |
| LabelAndPrediction(labels[i], |
| adjusted_predictor([sensitive_features[i]], [scores[i]]))) |
| return labels_and_predictions |
|
|
|
|
| def _format_as_list_of_lists(lst): |
| return [[item] for item in lst] |
|
|
