Sound-Dr/utils.py at master · ReML-AI/Sound-Dr · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
import numpy as np
from sklearn.metrics import f1_score, confusion_matrix, roc_auc_score, auc, precision_recall_curve, roc_curve, average_precision_score

## Define functions
def merge_feature(list_features):
    """
      Merge numpy array features
      Args:
        - list_features: list of numpy array features
                         :type: a list of numpy arrays
      Returns:
        - features: the concatenate numpy array along axis=1
                    :type: a numpy array
    """
    features = np.concatenate(list_features, axis=1)
    features = np.nan_to_num(features)
    features = np.clip(features, -np.finfo(np.float32).max, np.finfo(np.float32).max)
    return features

def compute_metrics(cfs_matrix):
    """
      Calculate common metrics based on the confusion matrix
      Args:
        - cfs_matrix: a sklearn confusion matrix
                      :type: a ndarray of shape (n_classes, n_classes)
      Returns:
        - precision: the precision of the prediction
                     :type: float
        - recall: the recall of the prediction
                  :type: float
        - f1: the f1-score of the prediction
              :type: float
    """
    precision = cfs_matrix[1,1] / (cfs_matrix[1,1] + cfs_matrix[0,1])
    recall = cfs_matrix[1,1] / (cfs_matrix[1,1] + cfs_matrix[1,0])
    f1 = 2 * (precision * recall) / (precision + recall)
    return precision, recall, f1

# # Evaluate
def evaluate(ensem_preds, targets):
    """
      Evaluate the prediction by providing metrics & also the best threshold (to get the highest f1-score)
      Ex: AUC, Accurary, Precision, Recall, F1-Score.
      Then print these metrics
      Args:
        - ensem_preds: predictions for ids
                       :type: a numpy array
        - targets: the actual results of ids
                   :type: a numpy array
      Returns:
        - None
    """
    best_th = 0
    best_score = 0

    for th in np.arange(0.0, 0.6, 0.01):
        pred = (ensem_preds > th).astype(int)
        score = f1_score(targets, pred)
        if score > best_score:
            best_th = th
            best_score = score

    print(f"\nAUC score: {roc_auc_score(targets, ensem_preds):12.4f}")
    print(f"Best threshold {best_th:12.4f}")

    preds = (ensem_preds > best_th).astype(int)

    cm1 = confusion_matrix(targets, preds)
    print('\nConfusion Matrix : \n', cm1)
    precision, recall, f1 = compute_metrics(cm1)

    print('\n=============')
    print (f'Precision    : {precision:12.4f}')

    print(f'Recall : {recall:12.4f}')

    print(f'F1 Score : {f1:12.4f}')

    total1=sum(sum(cm1))

    print('\n=============')
    accuracy1=(cm1[0,0]+cm1[1,1])/total1
    print (f'Accuracy    : {accuracy1:12.4f}')