-
Notifications
You must be signed in to change notification settings - Fork 0
/
classification-analysis.py
194 lines (157 loc) · 6.97 KB
/
classification-analysis.py
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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
# ==============================================================================
# THIS FILE PERFORMS THE CLASSIFICATION USING MBI AND MFI STATISTICS
#
# ==============================================================================
import pandas as pd
import numpy as np
from numpy.matlib import repmat
import matplotlib.pyplot as plt
from scipy import interp
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import StratifiedKFold
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix
from sklearn.metrics import roc_curve
from sklearn.metrics import roc_auc_score
from sklearn.metrics import recall_score
from sklearn.preprocessing import LabelBinarizer, OneHotEncoder
# ==============================================================================
# SET SOME PARAMETERS
# ==============================================================================
nsteps = 10
rng = np.random.RandomState(6634)
rng_lrc = np.random.RandomState(7432)
# ==============================================================================
# IMPORT DATA
# X = ndarray((nsubjects, 20))
# where X[,:10] are MBI values for 10 trials back
# and where X[:,10:] are MFI values for 10 trials back
# y = ndarray((nsubjects, 3))
# each row is a one hot vector [class=hc, class=bed, class=bn] for
# the given subject
# ==============================================================================
X = 'YOUR FEATURES HERE'
y_hot = 'YOUR ONE HOT CLASS LABELS HERE'
# ==============================================================================
# CLASSIFICATION
# ==============================================================================
c = 1
alpha = 0
lr_penalty='l1'
n_splits = int(N/4)
skf = StratifiedKFold(n_splits=n_splits)
AUC = []
COEFS_BN = []
mcfig, ax = plt.subplots(figsize=(6, 5))
ax.set_xlabel('False Positive Rate'); ax.set_xlim([0, 1])
ax.set_ylabel('True Positive Rate'); ax.set_ylim([0, 1])
ax.plot(np.linspace(0, 1, 100), np.linspace(0, 1, 100), ls='--', c='k', lw=1.5)
TPR_PBE = []
TPR_BED = []
TPR_BN = []
TPR_HCBD = []
TPR_HCBN = []
TPR_BDBN = []
mean_fpr = np.linspace(0, 1, 100)
for train_index, test_index in skf.split(X, np.argmax(y_hot, axis=1)):
X_train, X_test = X[train_index, :], X[test_index, :]
y_train, y_test = y_hot[train_index,:], y_hot[test_index,:]
scale_x = StandardScaler().fit(X_train)
X_train = scale_x.transform(X_train)
X_test = scale_x.transform(X_test)
# --------------------------------------------------------------------------
# INITIALIZE CLASSIFIERS
# --------------------------------------------------------------------------
lrc_pbe = LogisticRegression(random_state=rng_lrc, penalty=lr_penalty, C=c)
lrc_bed = LogisticRegression(random_state=rng_lrc, penalty=lr_penalty, C=c)
lrc_bn = LogisticRegression(random_state=rng_lrc, penalty=lr_penalty, C=c)
# --------------------------------------------------------------------------
# TRAIN
# --------------------------------------------------------------------------
lrc_pbe.fit(X=X_train, y=y_train[:, 0])
lrc_bed.fit(X=X_train, y=y_train[:, 1])
lrc_bn.fit(X=X_train, y=y_train[:, 2])
COEFS_BN.append(lrc_bn.coef_)
# --------------------------------------------------------------------------
# TEST
# --------------------------------------------------------------------------
yhat_pbe = lrc_pbe.predict_proba(X=X_test)
yhat_bed = lrc_bed.predict_proba(X=X_test)
yhat_bn = lrc_bn.predict_proba(X=X_test)
# --------------------------------------------------------------------------
# MEASURE PERFORMANCE
# --------------------------------------------------------------------------
auc_pbe = roc_auc_score(y_true=y_test[:, 0], y_score=yhat_pbe[:,1])
auc_bed = roc_auc_score(y_true=y_test[:, 1], y_score=yhat_bed[:,1])
auc_bn = roc_auc_score(y_true=y_test[:, 2], y_score=yhat_bn[:,1])
fpr_pbe, tpr_pbe, thresh_pbe = roc_curve(y_true=y_test[:, 0],
y_score=yhat_pbe[:,1])
fpr_bed, tpr_bed, thresh_bed = roc_curve(y_true=y_test[:, 1],
y_score=yhat_bed[:,1])
fpr_bn, tpr_bn, thresh_bn = roc_curve(y_true=y_test[:, 2],
y_score=yhat_bn[:,1])
TPR_PBE.append(interp(mean_fpr, fpr_pbe, tpr_pbe))
TPR_BED.append(interp(mean_fpr, fpr_bed, tpr_bed))
TPR_BN.append(interp(mean_fpr, fpr_bn, tpr_bn))
ax.plot(fpr_pbe, tpr_pbe, c='b', alpha=alpha, lw=1.5)
ax.plot(fpr_bed, tpr_bed, c='r', alpha=alpha, lw=1.5)
ax.plot(fpr_bn, tpr_bn, c='g', alpha=alpha, lw=1.5)
AUC.append([auc_pbe, auc_bed, auc_bn])
print('=========== AUC ===========')
AUC = np.array(AUC)
aucmean = np.round(np.mean(AUC, axis=0), 2)
aucstd = np.std(AUC, axis=0)
auclci = np.round(aucmean - 1.96*aucstd/np.sqrt(AUC.shape[0]), 2)
aucuci = np.round(aucmean + 1.96*aucstd/np.sqrt(AUC.shape[0]), 2)
# FINALIZE ROC CURVES
mean_tpr_pbe = np.mean(TPR_PBE, axis=0); mean_tpr_pbe[-1] = 1; mean_tpr_pbe[0]=0
mean_tpr_bed = np.mean(TPR_BED, axis=0); mean_tpr_bed[-1] = 1; mean_tpr_bed[0]=0
mean_tpr_bn = np.mean(TPR_BN, axis=0); mean_tpr_bn[-1] = 1; mean_tpr_bn[0]=0
ax.plot(mean_fpr, mean_tpr_pbe,
c='#374E55FF', lw=2,
label='HC-Rest AUC=' +
str(aucmean[0]) + ' (' +
str(auclci[0]) + '-' + str(aucuci[0]) + ')')
ax.plot(mean_fpr, mean_tpr_bed,
c='#DF8F44FF', lw=2,
label='BED-Rest AUC='+
str(aucmean[1]) + ' (' +
str(auclci[1]) + '-' + str(aucuci[1]) + ')')
ax.plot(mean_fpr, mean_tpr_bn,
c='#00A1D5FF', lw=2,
label='BN-Rest AUC='+
str(aucmean[2]) + ' (' +
str(auclci[2]) + '-' + str(aucuci[2]) + ')')
plt.legend()
plt.savefig('results/plots/look10back-prediction-cv.png',
bbox_inches='tight',
dpi=300)
plt.show()
mean_coef_bn = np.mean(np.array(COEFS_BN), axis=0).flatten()
std_coef_bn = (np.std(np.array(COEFS_BN),axis=0)/np.sqrt(n_splits)).flatten()
fig, ax = plt.subplots(nrows=2, figsize=(8, 4))
ax[0].set_title('Model-Based Index')
ax[0].bar(np.arange(10)+1,
mean_coef_bn[:10],
yerr=std_coef_bn[:10],
color='#374E55FF',
edgecolor='k',
linewidth=1.5)
ax[0].set_ylabel('Coefficient')
ax[0].set_xticklabels([])
ax[0].set_xticks([])
ax[1].set_title('Model-Free Index')
ax[1].bar(np.arange(10)+1,
mean_coef_bn[10:],
yerr=std_coef_bn[10:],
color='#DF8F44FF',
edgecolor='k',
linewidth=1.5)
ax[1].set_ylabel('Coefficient')
ax[1].set_xticks(np.arange(10)+1)
ax[1].set_xlabel('N Steps Back')
plt.tight_layout()
plt.savefig('results/plots/importance-lasso-bn.png',
bbox_inches='tight',
dpi=300)
plt.show()