def gender_performance(df_test_encoded, predictions, print_=False):
predictions_m = []
predictions_f = []
df_test_encoded_m = []
df_test_encoded_f = []
tamanho = len(df_test_encoded['sex'])
for i in range(tamanho):
if df_test_encoded['sex'].iloc[i] == 1:
df_test_encoded_m.append(df_test_encoded['earnings'].iloc[i])
predictions_m.append(predictions[i])
else:
df_test_encoded_f.append(df_test_encoded['earnings'].iloc[i])
predictions_f.append(predictions[i])
true_positive_m = true_positive_rate(df_test_encoded_m, predictions_m, pos_label=1)
false_positive_m = false_positive_rate(df_test_encoded_m, predictions_m, pos_label=1)
true_negative_m = true_negative_rate(df_test_encoded_m, predictions_m, pos_label=1)
false_negative_m = false_negative_rate(df_test_encoded_m, predictions_m, pos_label=1)
true_positive_f = true_positive_rate(df_test_encoded_f, predictions_f, pos_label=1)
false_positive_f = false_positive_rate(df_test_encoded_f, predictions_f, pos_label=1)
true_negative_f = true_negative_rate(df_test_encoded_f, predictions_f, pos_label=1)
false_negative_f = false_negative_rate(df_test_encoded_f, predictions_f, pos_label=1)
if print_:
print("True Positive Rate for Male:", true_positive_m)
print("True Positive Rate for Female:", true_positive_f)
print("False Positive Rate for Male:", false_positive_m)
print("False Positive Rate for Female:", false_positive_f)
print("True Negative Rate for Male:", true_negative_m)
print("True Negative Rate for Female:", true_negative_f)
print("False Negative Rate for Male:", false_negative_m)
print("False Negative Rate for Female:", false_negative_f)
def test_some_correct(self):
y_true = [0, 0, 0, 0, 1, 1, 1, 1]
y_pred = [0, 0, 0, 1, 1, 0, 0, 0]
result = metrics.true_positive_rate(y_true, y_pred)
assert result == 0.25
result = metrics.true_positive_rate(y_true, y_pred, pos_label=0)
assert result == 0.75
def test_all_correct(self):
y_true = [0, 0, 0, 0, 1]
y_pred = [0, 0, 0, 0, 1]
result = metrics.true_positive_rate(y_true, y_pred)
assert result == 1
result = metrics.true_positive_rate(y_true, y_pred, pos_label=0)
assert result == 1
def test_against_sklearn(self):
y_true = [0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1]
y_pred = [1, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1]
result = metrics.true_positive_rate(y_true, y_pred)
result_skm = skm.recall_score(y_true, y_pred)
assert result == pytest.approx(result_skm)
result = metrics.true_positive_rate(y_true, y_pred, pos_label=0)
result_skm = skm.recall_score(y_true, y_pred, pos_label=0)
assert result == pytest.approx(result_skm)
def test_against_sklearn_weighted(self):
y_true = [0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1]
y_pred = [1, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1]
weight = [1, 2, 4, 5, 1, 2, 2, 3, 1, 2, 1, 3, 1, 5, 4, 2]
result = metrics.true_positive_rate(y_true, y_pred, sample_weight=weight)
result_skm = skm.recall_score(y_true, y_pred, sample_weight=weight)
assert result == pytest.approx(result_skm)
result = metrics.true_positive_rate(y_true, y_pred,
pos_label=0, sample_weight=weight)
result_skm = skm.recall_score(y_true, y_pred,
pos_label=0, sample_weight=weight)
assert result == pytest.approx(result_skm)
def test_tpr_values_alpha(self):
a = 'a'
b = 'b'
y_true = [a, b, a, b, b, b, b, b, a, a, a, a, b, b, a, b]
y_pred = [a, b, b, a, a, b, a, a, b, b, a, b, a, b, a, a]
# Use sklearn for validation
result_a = metrics.true_positive_rate(y_true, y_pred, pos_label=a)
result_skm_a = skm.recall_score(y_true, y_pred, pos_label=a)
assert result_a == pytest.approx(result_skm_a)
result_b = metrics.true_positive_rate(y_true, y_pred, pos_label=b)
result_skm_b = skm.recall_score(y_true, y_pred, pos_label=b)
assert result_b == pytest.approx(result_skm_b)
# Sanity check on the in put data
assert result_a != result_b
def test_all_negative_ones(self):
neg_ones = -1 * np.ones(10)
# Should behave as if pos_label=1
assert metrics.true_positive_rate(neg_ones, neg_ones) == 0
assert metrics.false_positive_rate(neg_ones, neg_ones) == 0
assert metrics.true_negative_rate(neg_ones, neg_ones) == 1
assert metrics.false_negative_rate(neg_ones, neg_ones) == 0
assert metrics.true_positive_rate(neg_ones, neg_ones, pos_label=1) == 0
assert metrics.false_positive_rate(neg_ones, neg_ones, pos_label=1) == 0
assert metrics.true_negative_rate(neg_ones, neg_ones, pos_label=1) == 1
assert metrics.false_negative_rate(neg_ones, neg_ones, pos_label=1) == 0
assert metrics.true_positive_rate(neg_ones, neg_ones, pos_label=-1) == 1
assert metrics.false_positive_rate(neg_ones, neg_ones, pos_label=-1) == 0
assert metrics.true_negative_rate(neg_ones, neg_ones, pos_label=-1) == 0
assert metrics.false_negative_rate(neg_ones, neg_ones, pos_label=-1) == 0
def test_all_ones(self):
ones = np.ones(10)
# should behave as if pos_label=1
assert metrics.true_positive_rate(ones, ones) == 1
assert metrics.false_positive_rate(ones, ones) == 0
assert metrics.true_negative_rate(ones, ones) == 0
assert metrics.false_negative_rate(ones, ones) == 0
assert metrics.true_positive_rate(ones, ones, pos_label=1) == 1
assert metrics.false_positive_rate(ones, ones, pos_label=1) == 0
assert metrics.true_negative_rate(ones, ones, pos_label=1) == 0
assert metrics.false_negative_rate(ones, ones, pos_label=1) == 0
assert metrics.true_positive_rate(ones, ones, pos_label=0) == 0
assert metrics.false_positive_rate(ones, ones, pos_label=0) == 0
assert metrics.true_negative_rate(ones, ones, pos_label=0) == 1
assert metrics.false_negative_rate(ones, ones, pos_label=0) == 0
def test_all_zeros(self):
zeros = np.zeros(10)
# Should behave as if pos_label=1
assert metrics.true_positive_rate(zeros, zeros) == 0
assert metrics.false_positive_rate(zeros, zeros) == 0
assert metrics.true_negative_rate(zeros, zeros) == 1
assert metrics.false_negative_rate(zeros, zeros) == 0
assert metrics.true_positive_rate(zeros, zeros, pos_label=1) == 0
assert metrics.false_positive_rate(zeros, zeros, pos_label=1) == 0
assert metrics.true_negative_rate(zeros, zeros, pos_label=1) == 1
assert metrics.false_negative_rate(zeros, zeros, pos_label=1) == 0
assert metrics.true_positive_rate(zeros, zeros, pos_label=0) == 1
assert metrics.false_positive_rate(zeros, zeros, pos_label=0) == 0
assert metrics.true_negative_rate(zeros, zeros, pos_label=0) == 0
assert metrics.false_negative_rate(zeros, zeros, pos_label=0) == 0
