7 examples of 'how to calculate auc manually' in Python

654
def calc_auc(x, y):
655
    """ Given x and y values it calculates the approx. integral and normalizes it: area under curve"""
656
    integral = np.trapz(y, x)
657
    norm = np.trapz(np.ones_like(y), x)
658
659
    return integral / norm

363
def auc(self):
364
    if self.type != DatasetType.binary:
365
        # raise ValueError("AUC metric is only supported for binary classification: {}.".format(self.classes))
366
        log.warning("AUC metric is only supported for binary classification: %s.", self.classes)
367
        return nan
368
    return float(roc_auc_score(self.truth, self.probabilities[:, 1]))

19
def auc_score(y_true, y_pred, positive_label=1):
20
    if hasattr(sklearn.metrics, 'roc_auc_score'):
21
        return sklearn.metrics.roc_auc_score(y_true, y_pred)
22
23
    fp_rate, tp_rate, thresholds = sklearn.metrics.roc_curve(
24
        y_true, y_pred, pos_label=positive_label)
25
    return sklearn.metrics.auc(fp_rate, tp_rate)

98
def plot_auc(self):
99
    if self.n_classes != 2:
100
        display("plot_auc() not yet implemented for multiclass classifiers")
101
        return None
102
103
    # Move binarized to classifier
104
    y_true_binarized = label_binarize(self.y_true, classes=self.classes)
105
    y_pred_binarized = 1 - self.y_pred_proba
106
107
    y_true_binarized = np.hstack((y_true_binarized, 1 - y_true_binarized))
108
    y_pred_binarized = np.hstack((y_pred_binarized, 1 - y_pred_binarized))
109
110
    fig = plt.figure()
111
112
    fpr = dict()
113
    tpr = dict()
114
    roc_auc = dict()
115
    for i in range(self.n_classes):
116
        fpr[i], tpr[i], _ = sklearn.metrics.roc_curve(
117
            y_true_binarized[:, i], y_pred_binarized[:, i]
118
        )
119
        roc_auc[i] = sklearn.metrics.auc(fpr[i], tpr[i])
120
121
        # return roc_auc
122
        self._plot_auc_label(fig, fpr[i], tpr[i], roc_auc[i], i)
123
124
    display(HTML("<h2>AUC Plot</h2>"))
125
    display(fig)

59
def _auc_arr(score):
60
    score_p = score[:,0]
61
    score_n = score[:,1]
62
63
    score_arr = []
64
    for s in score_p.tolist():
65
        score_arr.append([0,1,s])
66
    for s in score_n.tolist():
67
        score_arr.append([1,0,s])
68
    return score_arr

190
def compute_negative_cross_auc(df, subgroup, label, model_name):
191
  """Computes the AUC of the within-subgroup negative examples and the background positive examples."""
192
  subgroup_negative_examples = df[df[subgroup] & ~df[label]]
193
  non_subgroup_positive_examples = df[~df[subgroup] & df[label]]
194
  examples = subgroup_negative_examples.append(non_subgroup_positive_examples)
195
  return compute_auc(examples[label], examples[model_name])

111
def calc_metrics(testy, scores):
112
    precision, recall, _ = precision_recall_curve(testy, scores)
113
    roc_auc = roc_auc_score(testy, scores)
114
    prc_auc = auc(recall, precision)
115
116
    return roc_auc, prc_auc