bme6525-project/methods-comparison.py at main · avzuben/bme6525-project · 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
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

from scipy.stats import mode
from sklearn.metrics import confusion_matrix

from utils import get_df, pathologies

ifig = 0


def plot_confusion_matrix(cm, title, class_labels=None):
    global ifig
    ifig += 1
    plt.figure(ifig)
    plt.title(title)
    sns.heatmap(cm, annot=True, linewidths=.1, fmt= '.0f', cmap='Reds')
    plt.xlabel('Predicted')
    plt.ylabel('Actual')
    if class_labels is not None:
        plt.xticks(np.arange(len(class_labels)) + 0.5, class_labels)
        plt.yticks(np.arange(len(class_labels)) + 0.5, class_labels)
    plt.show()


df = get_df()
y = df[df.columns[-1]]

rf_pred = np.load('./random_forest_predictions.npy', allow_pickle=True)
rf_acc = np.mean(rf_pred == y)
cm = confusion_matrix(y, rf_pred)
rf_confidence = cm / np.sum(cm, 0)
rf_conf_pred = np.array([rf_confidence[:, int(pred)] for pred in rf_pred])
rf_w_pred = np.zeros((len(rf_pred), len(pathologies.keys())))
for i, pred in enumerate(rf_pred):
    rf_w_pred[i, int(pred)] = rf_acc
plot_confusion_matrix(cm, 'Random Forest - Confusion Matrix', pathologies.keys())

ls_pred = np.load('./linear_svm_predictions.npy', allow_pickle=True)
ls_acc = np.mean(ls_pred == y)
cm = confusion_matrix(y, ls_pred)
ls_confidence = cm / np.sum(cm, 0)
ls_conf_pred = np.array([ls_confidence[:, int(pred)] for pred in rf_pred])
ls_w_pred = np.zeros((len(ls_pred), len(pathologies.keys())))
for i, pred in enumerate(ls_pred):
    ls_w_pred[i, int(pred)] = ls_acc
plot_confusion_matrix(cm, 'Linear SVM - Confusion Matrix', pathologies.keys())

qs_pred = np.load('./quadratic_svm_predictions.npy', allow_pickle=True)
qs_acc = np.mean(qs_pred == y)
cm = confusion_matrix(y, qs_pred)
qs_confidence = cm / np.sum(cm, 0)
qs_conf_pred = np.array([qs_confidence[:, int(pred)] for pred in rf_pred])
qs_w_pred = np.zeros((len(qs_pred), len(pathologies.keys())))
for i, pred in enumerate(qs_pred):
    qs_w_pred[i, int(pred)] = qs_acc
plot_confusion_matrix(cm, 'Quadratic SVM - Confusion Matrix', pathologies.keys())

rs_pred = np.load('./rbf_svm_predictions.npy', allow_pickle=True)
rs_acc = np.mean(rs_pred == y)
cm = confusion_matrix(y, rs_pred)
rs_confidence = cm / np.sum(cm, 0)
rs_w_pred = np.zeros((len(rs_pred), len(pathologies.keys())))
for i, pred in enumerate(rs_pred):
    rs_w_pred[i, int(pred)] = rs_acc
rs_conf_pred = np.array([rs_confidence[:, int(pred)] for pred in rf_pred])
plot_confusion_matrix(cm, 'RBF SVM - Confusion Matrix', pathologies.keys())

nn_pred = np.load('./mlp_predictions.npy', allow_pickle=True)
nn_acc = np.mean(nn_pred == y)
cm = confusion_matrix(y, nn_pred)
nn_confidence = cm / np.sum(cm, 0)
nn_w_pred = np.zeros((len(nn_pred), len(pathologies.keys())))
for i, pred in enumerate(nn_pred):
    nn_w_pred[i, int(pred)] = nn_acc
nn_conf_pred = np.array([nn_confidence[:, int(pred)] for pred in rf_pred])
plot_confusion_matrix(cm, 'MLP - Confusion Matrix', pathologies.keys())

vote = np.array(mode(np.concatenate([[rf_pred], [ls_pred], [qs_pred], [rs_pred], [nn_pred]])))[0].flatten()
vote_acc = np.mean(vote == y)
cm = confusion_matrix(y, vote)
plot_confusion_matrix(cm, 'Vote - Confusion Matrix', pathologies.keys())

top3_vote = np.array(mode(np.concatenate([[rf_pred], [rs_pred], [nn_pred]])))[0].flatten()
top3_vote_acc = np.mean(top3_vote == y)
cm = confusion_matrix(y, top3_vote)
plot_confusion_matrix(cm, 'Top-3 Vote - Confusion Matrix', pathologies.keys())

rf_seg_pred = np.load('./random_forest_seg_predictions.npy', allow_pickle=True)
rf_seg_acc = np.mean(rf_seg_pred == y)
rf_seg_conf_pred = np.array([rf_confidence[:, int(pred)] for pred in rf_seg_pred])
rf_seg_w_pred = np.zeros((len(rf_seg_pred), len(pathologies.keys())))
for i, pred in enumerate(rf_seg_pred):
    rf_seg_w_pred[i, int(pred)] = rf_acc
cm = confusion_matrix(y, rf_seg_pred)
plot_confusion_matrix(cm, 'Random Forest - Segmentation - Confusion Matrix', pathologies.keys())

ls_seg_pred = np.load('./linear_seg_svm_predictions.npy', allow_pickle=True)
ls_seg_acc = np.mean(ls_seg_pred == y)
ls_seg_conf_pred = np.array([ls_confidence[:, int(pred)] for pred in ls_seg_pred])
ls_seg_w_pred = np.zeros((len(ls_seg_pred), len(pathologies.keys())))
for i, pred in enumerate(ls_seg_pred):
    ls_seg_w_pred[i, int(pred)] = ls_acc
cm = confusion_matrix(y, ls_seg_pred)
plot_confusion_matrix(cm, 'Linear SVM - Segmentation - Confusion Matrix', pathologies.keys())

qs_seg_pred = np.load('./quadratic_seg_svm_predictions.npy', allow_pickle=True)
qs_seg_acc = np.mean(qs_seg_pred == y)
qs_seg_conf_pred = np.array([qs_confidence[:, int(pred)] for pred in qs_seg_pred])
qs_seg_w_pred = np.zeros((len(qs_seg_pred), len(pathologies.keys())))
for i, pred in enumerate(qs_seg_pred):
    qs_seg_w_pred[i, int(pred)] = qs_acc
cm = confusion_matrix(y, qs_seg_pred)
plot_confusion_matrix(cm, 'Quadratic SVM - Segmentation - Confusion Matrix', pathologies.keys())

rs_seg_pred = np.load('./rbf_seg_svm_predictions.npy', allow_pickle=True)
rs_seg_acc = np.mean(rs_seg_pred == y)
rs_seg_conf_pred = np.array([rs_confidence[:, int(pred)] for pred in rs_seg_pred])
rs_seg_w_pred = np.zeros((len(rs_seg_pred), len(pathologies.keys())))
for i, pred in enumerate(rs_seg_pred):
    rs_seg_w_pred[i, int(pred)] = rs_acc
cm = confusion_matrix(y, rs_seg_pred)
plot_confusion_matrix(cm, 'RBF SVM - Segmentation - Confusion Matrix', pathologies.keys())

nn_seg_pred = np.load('./mlp_seg_predictions.npy', allow_pickle=True)
nn_seg_acc = np.mean(nn_seg_pred == y)
nn_seg_conf_pred = np.array([nn_confidence[:, int(pred)] for pred in nn_seg_pred])
nn_seg_w_pred = np.zeros((len(nn_seg_pred), len(pathologies.keys())))
for i, pred in enumerate(nn_seg_pred):
    nn_seg_w_pred[i, int(pred)] = nn_acc
cm = confusion_matrix(y, nn_seg_pred)
plot_confusion_matrix(cm, 'MLP - Segmentation - Confusion Matrix', pathologies.keys())

vote_seg = np.array(mode(np.concatenate([[rf_seg_pred], [ls_seg_pred], [qs_seg_pred], [rs_seg_pred], [nn_seg_pred]])))[0].flatten()
vote_seg_acc = np.mean(vote_seg == y)
cm = confusion_matrix(y, vote_seg)
plot_confusion_matrix(cm, 'Vote - Segmentation - Confusion Matrix', pathologies.keys())

top3_vote_seg = np.array(mode(np.concatenate([[rf_seg_pred], [rs_seg_pred], [nn_seg_pred]])))[0].flatten()
top3_vote_seg_acc = np.mean(top3_vote_seg == y)
cm = confusion_matrix(y, top3_vote_seg)
plot_confusion_matrix(cm, 'Top-3 Vote - Segmentation - Confusion Matrix', pathologies.keys())

cnn_pred = np.load('./cnn_predictions.npy', allow_pickle=True)
cnn_acc = np.mean(np.argmax(cnn_pred, -1) == y)
cm = confusion_matrix(y, np.argmax(cnn_pred, -1))
plot_confusion_matrix(cm, 'CNN - Confusion Matrix', pathologies.keys())

print('Accuracies')
print('Random Forest:', rf_acc)
print('Linear SVM:', ls_acc)
print('Quadratic SVM:', qs_acc)
print('RBF SVM:', rs_acc)
print('MLP:', nn_acc)
print('Vote:', vote_acc)
print('Top-3 Vote:', top3_vote_acc)

print(' ')
print('Segmentation Accuracies')
print('Random Forest:', rf_seg_acc)
print('Linear SVM:', ls_seg_acc)
print('Quadratic SVM:', qs_seg_acc)
print('RBF SVM:', rs_seg_acc)
print('MLP:', nn_seg_acc)
print('Vote:', vote_seg_acc)
print('Top-3 Vote:', top3_vote_seg_acc)

print(' ')
print('CNN Accuracies')
print('CNN:', cnn_acc)