SGNNMD/utils.py at main · bubblecode/SGNNMD · 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
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
from sklearn import metrics
import time
import numpy as np
import matplotlib.pyplot as plt
import matplotlib
matplotlib.use('Agg')


def draw_acc_loss(train_acc, train_loss, val_acc, val_loss, path=None, title=None):
    epoch = list(range(len(train_acc)))
    if path is None:
        path = 'train_loss__{}.png'.format(str(time.time()))
    if title is None:
        title = 'undefined'
    plt.cla()
    plt.plot(epoch, train_loss, 'b-', label='train_loss')
    plt.plot(epoch, val_loss, 'k-', label='val_loss')
    plt.plot(epoch, train_acc, 'r-', label='train_acc')
    plt.plot(epoch, val_acc, 'g-', label='val_acc')
    plt.title(title)
    plt.legend()
    plt.savefig(path)


# def getTopMetrics(preds, targets):
#     ## TOP: 取值高的那一个进行指标计算
#     """
#     请确保传入的数据已做过如下操作：output.argmax(axis=1).tolist()
#     preds:   [0.333, 0.444, 0.555, ...]
#     targets：[1, 0, 1, 0, 1, ...]
#     """
#     preds = np.array(preds)
#     targets = np.array(targets)
#     all_targets = np.array(targets)
#     fpr, tpr, _ = metrics.roc_curve(targets, preds)
#     auc = metrics.auc(fpr, tpr)
#     pr, re, _ = metrics.precision_recall_curve(targets, preds)
#     aupr = metrics.auc(re, pr)
#     # preds[np.where(preds >= 0.5)] = 1
#     # preds[np.where(preds < 0.5)] = 0
#     precision = metrics.precision_score(targets, preds)
#     recall = metrics.recall_score(targets, preds)
#     f1 = metrics.f1_score(targets, preds)
#     return [aupr, auc, precision, recall, f1]

def getTopMetrics(preds, targets):
    ## TOP: 取值高的那一个进行指标计算
    """
    请确保传入的数据已做过如下操作：output.argmax(axis=1).tolist()
    preds:   [0.333, 0.444, 0.555, ...]
    targets：[1, 0, 1, 0, 1, ...]
    """
    preds = np.array(preds)
    targets = np.array(targets)
    all_targets = np.array(targets)
    fpr, tpr, _ = metrics.roc_curve(targets, preds)
    auc = metrics.auc(fpr, tpr)
    pr, re, _ = metrics.precision_recall_curve(targets, preds)
    aupr = metrics.auc(re, pr)
    # preds[np.where(preds >= 0.5)] = 1
    # preds[np.where(preds < 0.5)] = 0
    precision = metrics.precision_score(targets, preds)
    recall = metrics.recall_score(targets, preds)
    f1 = metrics.f1_score(targets, preds)
    return [aupr, auc, precision, recall, f1]


def getMacroMetrics(predict_score, real_score):
    """
    output: np.matrix([[0.11,0.22,...]])
    target: np.matrix([[1,0,1,0,1,...]])
    """
    l = np.array(real_score)
    p = predict_score  ## 此处存在链接的概率
    l[np.where(l != 0)] = 1
    p[np.where(p > 0.05)]= 1; p = p.astype(np.int)
    fpr, tpr, _ = metrics.roc_curve(l, p)
    auc = metrics.auc(fpr, tpr)
    pr, re, _ = metrics.precision_recall_curve(l, p)
    aupr = metrics.auc(re, pr)
    precision = metrics.precision_score(l, p)
    recall = metrics.recall_score(l, p)
    f1 = metrics.f1_score(l, p)
    return [aupr, auc, f1, precision, recall]


# def calc_metrics(output, targets):
#     target = np.copy(targets)

#     # TODO 计算第一项指标
#     target[np.where(target == 1)[0]] = 2
#     target[np.where(target == 0)[0]] = 1
#     target[np.where(target == 2)[0]] = 0
#     fpr, tpr, _ = metrics.roc_curve(target, class1)
#     auc1 = metrics.auc(fpr, tpr)
#     pr, re, _ = metrics.precision_recall_curve(target, class1)
#     aupr1 = metrics.auc(re, pr)

#     # TODO 计算第二项指标
#     target = np.copy(targets)
#     target[np.where(target == 2)[0]] = 0
#     fpr, tpr, _ = metrics.roc_curve(target, class2)
#     auc2 = metrics.auc(fpr, tpr)
#     pr, re, _ = metrics.precision_recall_curve(target, class2)
#     aupr2 = metrics.auc(re, pr)

#     # TODO 计算第三项指标
#     target = np.copy(targets)
#     target[np.where(target == 1)[0]] = 0
#     target[np.where(target == 2)[0]] = 1
#     fpr, tpr, _ = metrics.roc_curve(target, class3)
#     auc3 = metrics.auc(fpr, tpr)
#     pr, re, _ = metrics.precision_recall_curve(target, class3)
#     aupr3 = metrics.auc(re, pr)

#     precision = metrics.precision_score(
#         targets, class_all.detach().numpy(), average="macro")
#     recall = metrics.recall_score(
#         targets, class_all.detach().numpy(), average="macro")
#     f1 = metrics.f1_score(targets, class_all.detach().numpy(), average="macro")
#     # 取平均返回
#     return (aupr1+aupr2+aupr3)/3, \
#            (auc1+auc2+auc3)/3, \
#         precision, \
#         recall, \
#         f1