def prob_tester():
# Get Files
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5' # use quantile-based MI threshold
winsize = '2'
state_switch = 's1' # select the seizure state
# randomState = 42 # fix random state
np.random.seed(42) # fix randomness
th = 90 # select the MI threshold
# SVC
# kern = 'linear'
# clf = SVC(kernel=kern, probability=True)
# Logistic Regression
pen = 'l2'
clf = LogisticRegression(penalty=pen)
# Random Forest
# numEsts = 100
# clf = RandomForestClassifier(n_estimators=numEsts)
# AdaBoost
# algo = "SAMME"
# numEsts = 100
# base_clf = DecisionTreeClassifier(max_depth=1)
# clf = AdaBoostClassifier(base_estimator=base_clf, n_estimators=numEsts, algorithm=algo)
# Train model
X_train, y_train, X_test, y_test = gmi_dataset_extract(ldir,
gmitype,
winsize,
th,
state_switch,
interTestSize=0.66)
clf.fit(X_train, y_train)
def test_fullseizextract():
ldir = get_mat_root() + 'mlv2/fullseiz/th/'
patient = 'DV'
seizure = '10'
mf = full_seizure_extract(ldir, patient, seizure, 90)
print('dummy line')
def test_fullseizextract():
ldir = get_mat_root() + 'mlv2/fullseiz/th/'
patient = 'DV'
seizure = '10'
mf = full_seizure_extract(ldir, patient, seizure, 90)
print('dummy line')
def prob_tester():
# Get Files
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5' # use quantile-based MI threshold
winsize = '2'
state_switch = 's1' # select the seizure state
# randomState = 42 # fix random state
np.random.seed(42) # fix randomness
th = 90 # select the MI threshold
# SVC
# kern = 'linear'
# clf = SVC(kernel=kern, probability=True)
# Logistic Regression
pen = 'l2'
clf = LogisticRegression(penalty=pen)
# Random Forest
# numEsts = 100
# clf = RandomForestClassifier(n_estimators=numEsts)
# AdaBoost
# algo = "SAMME"
# numEsts = 100
# base_clf = DecisionTreeClassifier(max_depth=1)
# clf = AdaBoostClassifier(base_estimator=base_clf, n_estimators=numEsts, algorithm=algo)
# Train model
X_train, y_train, X_test, y_test = gmi_dataset_extract(ldir, gmitype, winsize, th, state_switch, interTestSize=0.66)
clf.fit(X_train, y_train)
def rec_test_result(save_type, loc, in_dict):
saveloc = get_mat_root() + loc
ttl = save_type + '_' + datetime.datetime.now().strftime("%Y-%m-%d_%H%M%S")
# ttl = save_type + '_' + datetime.datetime.now().strftime("%Y-%m-%d_") + append_string # need append_string as input
sv_dict = dict_cleaner(in_dict)
# print('Saving .mat file...')
sio.savemat(saveloc + ttl, sv_dict)
# print('Saving .mat file... Done.')
return None
def my_process(numEsts):
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5' # use quantile-based MI threshold
winsize = '2'
state_switch = 's1' # select the seizure state
inter_test_size = 0.66
max_windows = 9
# randomState = 42 # fix random state
np.random.seed(42) # fix randomness
th = 95 # select the MI threshold
clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1)
# AdaBoost
# algo = "SAMME"
# numEsts = 100
# base_clf = DecisionTreeClassifier(max_depth=1)
# clf = AdaBoostClassifier(base_estimator=base_clf, n_estimators=numEsts, algorithm=algo)
X_train, y_train, X_test, y_test = gmi_dataset_extract(
ldir, gmitype, winsize, th, state_switch, inter_test_size, max_windows)
clf.fit(X_train, y_train)
probas_ = clf.predict_proba(X_test)
fpr, tpr, thresholds = roc_curve(y_test, probas_[:, 1])
roc_auc = auc(fpr, tpr)
savedir = "mlv2/rocbin/num_est/"
savetype = "TEST"
comment = dict()
comment['ClassifierType'] = clf.__class__.__name__
comment['MaxWindows'] = float(max_windows % 5)
if "RandomForest" in comment['ClassifierType']:
comment['number_estimators'] = numEsts
elif "AdaBoost" in comment['ClassifierType']:
comment['BaseClassifier'] = base_clf.__class__.__name__
comment['number_estimators'] = numEsts
comment['algorithm'] = algo
elif "LogisticRegression" in comment['ClassifierType']:
comment['penalty'] = pen
elif "SVC" in comment['ClassifierType']:
comment['kernel'] = kern
comment['StateSwitch'] = state_switch
appendString = ''.join([state_switch, str(numEsts)])
rec_test_result(
savetype, savedir, {
'comment': comment,
'fpr': fpr,
'th': th,
'tpr': tpr,
'rocth': thresholds,
'rocauc': roc_auc
}, appendString)
def my_process(numEsts):
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5'
winsize = '2'
state_switch = 's1'
inter_test_size = 0.66
max_windows = 9
np.random.seed(42) # fix randomness
clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1)
print(clf.__class__.__name__ + " Threshold %0.0d/100" % th)
# get seizure epochs from each patient with predetermined training/testing division
X_train, y_train, X_test, y_test = gmi_dataset_extract(
ldir, gmitype, winsize, th, state_switch, inter_test_size, max_windows)
cv = StratifiedKFold(y_train, n_folds=5)
mean_tpr = 0.0
mean_fpr = np.linspace(0, 1, 100)
all_tpr = []
for i, (train, test) in enumerate(cv):
probas_ = clf.fit(X_train[train],
y_train[train]).predict_proba(X_train[test])
fpr, tpr, thresholds = roc_curve(y_train[test], probas_[:, 1])
mean_tpr += interp(mean_fpr, fpr, tpr)
mean_tpr[0] = 0.0
roc_auc = auc(fpr, tpr)
# plt.plot(fpr, tpr, lw=1, label='ROC fold %d (area = %0.2f)' % (i, roc_auc))
# plt.plot([0, 1], [0, 1], '--', color=(0.6, 0.6, 0.6), label='Luck')
mean_tpr /= len(cv)
mean_tpr[-1] = 1.0
score = auc(mean_fpr, mean_tpr)
# plt.plot(mean_fpr, mean_tpr, 'k--', label='Mean ROC (area = %0.2f)' % score, lw=2)
# PLOTTING
# plt.xlim([-0.05, 1.05])
# plt.ylim([-0.05, 1.05])
# plt.xlabel('False Positive Rate')
# plt.ylabel('True Positive Rate')
# plt.title('Receiver operating characteristic example')
# plt.legend(loc="lower right")
# plt.show()
# score = cross_val_score(clf, X_train, y_train.ravel(), cv=5)
# print("Accuracy: %0.4f (+/- %0.4f)" % (100*score.mean(), 100*score.std() * 2))
print(score)
scores.append(score)
def my_process(numEsts):
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5'
winsize = '2'
state_switch = 's1'
inter_test_size = 0.66
max_windows = 9
np.random.seed(42) # fix randomness
clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1)
print(clf.__class__.__name__ + " Threshold %0.0d/100" % th)
# get seizure epochs from each patient with predetermined training/testing division
X_train, y_train, X_test, y_test = gmi_dataset_extract(ldir, gmitype, winsize, th, state_switch, inter_test_size, max_windows)
cv = StratifiedKFold(y_train, n_folds=5)
mean_tpr = 0.0
mean_fpr = np.linspace(0, 1, 100)
all_tpr = []
for i, (train, test) in enumerate(cv):
probas_ = clf.fit(X_train[train], y_train[train]).predict_proba(X_train[test])
fpr, tpr, thresholds = roc_curve(y_train[test], probas_[:, 1])
mean_tpr += interp(mean_fpr, fpr, tpr)
mean_tpr[0] = 0.0
roc_auc = auc(fpr, tpr)
# plt.plot(fpr, tpr, lw=1, label='ROC fold %d (area = %0.2f)' % (i, roc_auc))
# plt.plot([0, 1], [0, 1], '--', color=(0.6, 0.6, 0.6), label='Luck')
mean_tpr /= len(cv)
mean_tpr[-1] = 1.0
score = auc(mean_fpr, mean_tpr)
# plt.plot(mean_fpr, mean_tpr, 'k--', label='Mean ROC (area = %0.2f)' % score, lw=2)
# PLOTTING
# plt.xlim([-0.05, 1.05])
# plt.ylim([-0.05, 1.05])
# plt.xlabel('False Positive Rate')
# plt.ylabel('True Positive Rate')
# plt.title('Receiver operating characteristic example')
# plt.legend(loc="lower right")
# plt.show()
# score = cross_val_score(clf, X_train, y_train.ravel(), cv=5)
# print("Accuracy: %0.4f (+/- %0.4f)" % (100*score.mean(), 100*score.std() * 2))
print(score)
scores.append(score)
def my_process(numEsts):
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5' # use quantile-based MI threshold
winsize = '2'
state_switch = 's1' # select the seizure state
inter_test_size = 0.66
max_windows = 9
# randomState = 42 # fix random state
np.random.seed(42) # fix randomness
th = 95 # select the MI threshold
clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1)
# AdaBoost
# algo = "SAMME"
# numEsts = 100
# base_clf = DecisionTreeClassifier(max_depth=1)
# clf = AdaBoostClassifier(base_estimator=base_clf, n_estimators=numEsts, algorithm=algo)
X_train, y_train, X_test, y_test = gmi_dataset_extract(ldir, gmitype, winsize, th, state_switch, inter_test_size, max_windows)
clf.fit(X_train, y_train)
probas_ = clf.predict_proba(X_test)
fpr, tpr, thresholds = roc_curve(y_test, probas_[:, 1])
roc_auc = auc(fpr, tpr)
savedir = "mlv2/rocbin/num_est/"
savetype = "TEST"
comment = dict()
comment['ClassifierType'] = clf.__class__.__name__
comment['MaxWindows'] = float(max_windows % 5)
if "RandomForest" in comment['ClassifierType']:
comment['number_estimators'] = numEsts
elif "AdaBoost" in comment['ClassifierType']:
comment['BaseClassifier'] = base_clf.__class__.__name__
comment['number_estimators'] = numEsts
comment['algorithm'] = algo
elif "LogisticRegression" in comment['ClassifierType']:
comment['penalty'] = pen
elif "SVC" in comment['ClassifierType']:
comment['kernel'] = kern
comment['StateSwitch'] = state_switch
appendString = ''.join([state_switch, str(numEsts)])
rec_test_result(savetype, savedir,
{'comment': comment, 'fpr': fpr, 'th': th, 'tpr': tpr, 'rocth': thresholds, 'rocauc': roc_auc},
appendString)
# import matplotlib.pyplot as plt from sklearn.cross_validation import StratifiedKFold from sklearn.metrics import auc, roc_curve from sklearn.ensemble import AdaBoostClassifier from sklearn.tree import DecisionTreeClassifier from sklearn.ensemble import RandomForestClassifier from sklearn.svm import SVC from sklearn.linear_model import LogisticRegression from scipy import interp from matransfer import gmi_dataset_extract from record_data import rec_test_result from get_root_dir import get_mat_root from joblib import Parallel, delayed import multiprocessing ldir = get_mat_root() + "mlv2/threshbin/" gmitype = 'gmi5' winsize = '2' state_switch = 's1' inter_test_size = 0.66 max_windows = 9 np.random.seed(42) # fix randomness # SVC # kern = 'linear' # clf = SVC(kernel=kern, probability=True) # Logistic Regression # pen = 'l2' # clf = LogisticRegression(penalty=pen) # Random Forest numEsts = 100 clf = RandomForestClassifier(n_estimators=numEsts)
import numpy as np # import matplotlib.pyplot as plt from sklearn.metrics import auc, roc_curve from sklearn.ensemble import AdaBoostClassifier, RandomForestClassifier from sklearn.tree import DecisionTreeClassifier from sklearn.svm import SVC from sklearn.linear_model import LogisticRegression # from sklearn.naive_bayes import BernoulliNB from matransfer import gmi_dataset_extract, full_seizure_extract, full_seizure_detect_save from get_root_dir import get_mat_root from record_data import rec_test_result # Get Files ldir = get_mat_root() + "mlv2/threshbin/" gmitype = 'gmi5' # use quantile-based MI threshold winsize = '2' state_switch = 's1' # select the seizure state # randomState = 42 # fix random state np.random.seed(42) # fix randomness th = 94 # select the MI threshold maxWindows = 10 # SVC # kern = 'linear' # clf = SVC(kernel=kern, probability=True) # Logistic Regression # pen = 'l2' # clf = LogisticRegression(penalty=pen) # Random Forest numEsts = 100 clf = RandomForestClassifier(n_estimators=numEsts) # AdaBoost
import numpy as np # import matplotlib.pyplot as plt from sklearn.metrics import auc, roc_curve from sklearn.ensemble import AdaBoostClassifier, RandomForestClassifier from sklearn.tree import DecisionTreeClassifier from sklearn.svm import SVC from sklearn.linear_model import LogisticRegression # from sklearn.naive_bayes import BernoulliNB from matransfer import gmi_dataset_extract from get_root_dir import get_mat_root from record_data import rec_test_result from joblib import Parallel, delayed import multiprocessing ldir = get_mat_root() + "mlv2/threshbin/" gmitype = 'gmi5' # use quantile-based MI threshold winsize = '2' state_switch = 's1' # select the seizure state inter_test_size = 0.66 max_windows = 10 np.random.seed(42) # fix randomness th = 95 # select the MI threshold # Random Forest numEsts = 100 clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1) # AdaBoost # algo = "SAMME" # numEsts = 100 # base_clf = DecisionTreeClassifier(max_depth=1) # clf = AdaBoostClassifier(base_estimator=base_clf, n_estimators=numEsts, algorithm=algo)
import numpy as np # import matplotlib.pyplot as plt from sklearn.metrics import auc, roc_curve from sklearn.ensemble import AdaBoostClassifier, RandomForestClassifier from sklearn.tree import DecisionTreeClassifier from sklearn.svm import SVC from sklearn.linear_model import LogisticRegression # from sklearn.naive_bayes import BernoulliNB from matransfer import gmi_dataset_extract, full_seizure_extract, full_seizure_detect_save from get_root_dir import get_mat_root from record_data import rec_test_result # Get Files ldir = get_mat_root() + "mlv2/threshbin/" gmitype = 'gmi5' # use quantile-based MI threshold winsize = '2' state_switch = 's2' # select the seizure state maxWindows = 10 # randomState = 42 # fix random state np.random.seed(42) # fix randomness th = 95 # SVC # kern = 'linear' # clf = SVC(kernel=kern, probability=True) # Logistic Regression # pen = 'l1' # clf = LogisticRegression(penalty=pen) # Random Forest numEsts = 100 clf = RandomForestClassifier(n_estimators=numEsts)
import numpy as np # import matplotlib.pyplot as plt from sklearn.metrics import auc, roc_curve from sklearn.ensemble import AdaBoostClassifier, RandomForestClassifier from sklearn.tree import DecisionTreeClassifier from sklearn.svm import SVC from sklearn.linear_model import LogisticRegression # from sklearn.naive_bayes import BernoulliNB from matransfer import gmi_dataset_extract, full_seizure_extract, full_seizure_detect_save from get_root_dir import get_mat_root from record_data import rec_test_result # Get Files ldir = get_mat_root() + "mlv2/threshbin/" gmitype = 'gmi5' # use quantile-based MI threshold winsize = '2' state_switch = 's1' # select the seizure state # randomState = 42 # fix random state np.random.seed(42) # fix randomness th = 94 # select the MI threshold maxWindows = 10 # SVC # kern = 'linear' # clf = SVC(kernel=kern, probability=True) # Logistic Regression # pen = 'l2' # clf = LogisticRegression(penalty=pen) # Random Forest numEsts = 100 clf = RandomForestClassifier(n_estimators=numEsts)
def my_process(th):
# th = 95 # for a fixed MI threshold
numEsts = 200 # for a fixed N_e
ldir = get_mat_root() + "mlv2/threshbin/"
gmitype = 'gmi5'
winsize = '2'
state_switch = 's2'
inter_test_size = 0.66
max_windows = 8
np.random.seed(42) # fix randomness
clf = RandomForestClassifier(n_estimators=numEsts, n_jobs=-1)
# print(clf.__class__.__name__ + " Threshold %0.0d/100" % th)
# get seizure epochs from each patient with predetermined training/testing division
X_train, y_train, X_test, y_test = gmi_dataset_extract(
ldir, gmitype, winsize, th, state_switch, inter_test_size, max_windows)
cv = StratifiedKFold(y_train, n_folds=5)
mean_tpr = 0.0
mean_fpr = np.linspace(0, 1, 100)
all_tpr = []
for i, (train, test) in enumerate(cv):
probas_ = clf.fit(X_train[train],
y_train[train]).predict_proba(X_train[test])
fpr, tpr, thresholds = roc_curve(y_train[test], probas_[:, 1])
mean_tpr += interp(mean_fpr, fpr, tpr)
mean_tpr[0] = 0.0
roc_auc = auc(fpr, tpr)
# plt.plot(fpr, tpr, lw=1, label='ROC fold %d (area = %0.2f)' % (i, roc_auc))
# plt.plot([0, 1], [0, 1], '--', color=(0.6, 0.6, 0.6), label='Luck')
mean_tpr /= len(cv)
mean_tpr[-1] = 1.0
score = auc(mean_fpr, mean_tpr)
# plt.plot(mean_fpr, mean_tpr, 'k--', label='Mean ROC (area = %0.2f)' % score, lw=2)
# PLOTTING
# plt.xlim([-0.05, 1.05])
# plt.ylim([-0.05, 1.05])
# plt.xlabel('False Positive Rate')
# plt.ylabel('True Positive Rate')
# plt.title('Receiver operating characteristic example')
# plt.legend(loc="lower right")
# plt.show()
# score = cross_val_score(clf, X_train, y_train.ravel(), cv=5)
# print("Accuracy: %0.4f (+/- %0.4f)" % (100*score.mean(), 100*score.std() * 2))
# print(score)
# scores.append(score)
# errs.append(score.std())
savedir = "mlv2/rocbin/num_est/th{}/".format(th)
savetype = "CV"
comment = dict()
comment['ClassifierType'] = clf.__class__.__name__
comment['MaxWindows'] = float(max_windows % 5)
if "RandomForest" in comment['ClassifierType']:
comment['number_estimators'] = numEsts
elif "AdaBoost" in comment['ClassifierType']:
comment['BaseClassifier'] = base_clf.__class__.__name__
comment['number_estimators'] = numEsts
comment['algorithm'] = algo
elif "LogisticRegression" in comment['ClassifierType']:
comment['penalty'] = pen
elif "SVC" in comment['ClassifierType']:
comment['kernel'] = kern
comment['StateSwitch'] = state_switch
appendString = ''.join([state_switch, str(numEsts)])
rec_test_result(savetype, savedir, {
'comment': comment,
'score': score,
'th': th
}, appendString)