def main():
CID = opts.cluster
if (opts.load != 'none'): CID = opts.load
X_train, X_test, Y_train, Y_test, X, X2, X3, enc = f.get_data_pro(
testsize=0.4)
test_accues = []
for n in range(1, 5):
model, history = md.bulid_model_nconv(X_train,
X_test,
Y_train,
Y_test,
X,
X2,
X3,
CID,
fromfile=opts.load,
n_dense=n)
test_accues.append(history.history['val_acc'][-1])
print()
print(test_accues)
# newData = X_test.reshape(X_test.shape[0], 1, 100, 20)
# Y_score = model.predict_proba(X_test)
# roc.roc_plot(
# Y_test, Y_score, 2, filepath=os.path.join('figures', CID + opts.title + 'roc.svg'),fmt='svg',title=opts.title)
# Y_de = decode_y(Y_test, features=enc.active_features_)
# Y_pred = model.predict(X_test)
# Y_depred = decode_y(Y_pred, features=enc.active_features_)
# print(classification_report(Y_de, Y_depred))
return
def main():
CID = opts.cluster
if (opts.load != 'none'): CID = opts.load
X_train, X_test, Y_train, Y_test, X, X2, X3, enc = f.get_data_pro(
testsize=0.2)
model = bulid_model(X_train,
X_test,
Y_train,
Y_test,
X,
X2,
X3,
CID,
fromfile=opts.load)
newData = X_test.reshape(X_test.shape[0], 1, 100, 20)
Y_score = model.predict_proba(X_test)
roc.roc_plot(Y_test,
Y_score,
2,
filepath=os.path.join('figures',
CID + opts.title + 'roc.svg'),
fmt='svg',
title=opts.title)
Y_de = decode_y(Y_test, features=enc.active_features_)
Y_pred = model.predict(X_test)
Y_depred = decode_y(Y_pred, features=enc.active_features_)
print(classification_report(Y_de, Y_depred))
return
def main():
CID = opts.cluster
if (opts.load != 'none'): CID = opts.load
X_train, X_test, Y_train, Y_test, X, X2, X3, enc = f.get_data_pro(
testsize=0)
X_shape = X_train.shape
X_holder = np.zeros(X_shape[0])
Y_inv = decode_y(Y_train)
print(X_shape)
skf = StratifiedKFold(n_splits=10)
skf.get_n_splits(X_train, Y_inv)
accues = []
aucs = []
for train_index, test_index in skf.split(X_train, Y_inv):
print("TRAIN:", train_index, "TEST:", test_index)
x_train, x_test = X_train[train_index], X_train[test_index]
y_train, y_test = Y_train[train_index], Y_train[test_index]
model = bulid_model(x_train,
x_test,
y_train,
y_test,
X,
X2,
X3,
CID,
fromfile=opts.load)
Y_de = decode_y(y_test, features=enc.active_features_)
Y_pred = model.predict(x_test)
Y_score = model.predict_proba(x_test)
fpr, tpr, th = roc_curve(Y_de, Y_score[:, 1])
Y_depred = decode_y(Y_pred, features=enc.active_features_)
accues.append(accuracy_score(Y_depred, Y_de))
aucs.append(auc(fpr, tpr))
print("###########################")
print(accues)
print(aucs)
# model, history = bulid_model(
# X_train, X_test, Y_train, Y_test, X, X2, X3, CID, fromfile=opts.load)
# #newData = X_test.reshape(X_test.shape[0], 1, 100, 20)
# Y_score = model.predict_proba(X_test)
# roc.roc_plot(
# Y_test,
# Y_score,
# 2,
# filepath=os.path.join('figures', CID + opts.title + 'roc.svg'),
# fmt='svg',
# title=opts.title)
# plt.close()
# print(history.history.keys())
# # summarize history for accuracy
# plt.plot(history.history['acc'])
# plt.plot(history.history['val_acc'])
# plt.title('model accuracy')
# plt.ylabel('accuracy')
# plt.xlabel('epoch')
# plt.legend(['train', 'test'], loc='upper left')
# plt.savefig(os.path.join('figures', CID + opts.title + 'learning-c.svg'),format='svg')
# plt.close()
#Y_de = decode_y(Y_test, features=enc.active_features_)
#Y_pred = model.predict(X_test)
#Y_depred = decode_y(Y_pred, features=enc.active_features_)
#print(classification_report(Y_de, Y_depred))
return
def main():
CID = opts.cluster
if (opts.load != 'none'): CID = opts.load
X_train, X_test, Y_train, Y_test, X, X2, X3, enc = f.get_data_pro(
testsize=0.4)
Y_inv = decode_y(Y_train)
Y_de_test = decode_y(Y_test)
ranges = np.linspace(.1, 1.0, 10)
test_accues = []
test_aucs = []
tt_acu_max = []
tt_acu_min = []
train_accues = []
train_aucs = []
tn_acu_max = []
tn_acu_min = []
for size in ranges:
x_train, x_placeholder, y_train, y_placeholder = train_test_split(
X_train, Y_train, test_size=1 - size, random_state=0)
#skf = StratifiedKFold(n_splits=10)
y_train_dec = decode_y(y_train, features=enc.active_features_)
model, history = bulid_model(x_train,
X_test,
y_train,
Y_test,
X,
X2,
X3,
CID,
fromfile=opts.load)
print('Poportion value')
print(history.history['val_acc'][-1])
print(history.history['acc'][-1])
#skf.get_n_splits(x_train, y_train_dec)
#accues = []
#aucs = []
Y_pred = model.predict(X_test)
Y_score = model.predict_proba(X_test)
Y_t_pred = model.predict(X_train)
Y_t_score = model.predict_proba(X_train)
fpr, tpr, thplaceholder = roc_curve(Y_de_test, Y_score[:, 1])
Y_depred = decode_y(Y_pred, features=enc.active_features_)
fpr_t, tpr_t, thplaceholder2 = roc_curve(Y_inv, Y_t_score[:, 1])
Y_t_depred = decode_y(Y_t_pred, features=enc.active_features_)
test_accues.append(history.history['val_acc'][-1])
tt_acu_max.append(np.max(history.history['val_acc']))
tt_acu_min.append(np.min(history.history['val_acc']))
train_accues.append(history.history['acc'][-1])
tn_acu_max.append(np.max(history.history['acc']))
tn_acu_min.append(np.min(history.history['acc']))
train_aucs.append(auc(fpr_t, tpr_t))
test_aucs.append(auc(fpr, tpr))
print("###########################")
print(test_accues)
print(test_aucs)
print(train_accues)
print(train_aucs)
print(classification_report(Y_de_test, Y_depred))
plt.figure()
plt.title("Learning cuvre to the model")
plt.xlabel("Training sample fraction")
plt.ylabel("Scores")
plt.grid()
plt.plot(np.linspace(.1, 1.0, 10),
train_accues,
'o-',
label="Training accurarcy")
plt.fill_between(np.linspace(.1, 1.0, 10),
tn_acu_min,
tn_acu_max,
alpha=0.1)
#plt.plot(np.linspace(.1, 1.0, 10),train_aucs,'o-',label="Training auc")
plt.plot(np.linspace(.1, 1.0, 10),
test_accues,
'o-',
label="Testing accurarcy")
plt.fill_between(np.linspace(.1, 1.0, 10),
tt_acu_min,
tt_acu_max,
alpha=0.1)
#plt.plot(np.linspace(.1, 1.0, 10),test_aucs,'o-',label="Testing auc")
plt.legend(loc="best")
plt.savefig(os.path.join('figures',
CID + opts.title + 'learning_curve.svg'),
format='svg')
plt.close()
# model, history = bulid_model(
# X_train, X_test, Y_train, Y_test, X, X2, X3, CID, fromfile=opts.load)
# #newData = X_test.reshape(X_test.shape[0], 1, 100, 20)
# Y_score = model.predict_proba(X_test)
# roc.roc_plot(
# Y_test,
# Y_score,
# 2,
# filepath=os.path.join('figures', CID + opts.title + 'roc.svg'),
# fmt='svg',
# title=opts.title)
# plt.close()
# print(history.history.keys())
# # summarize history for accuracy
# plt.plot(history.history['acc'])
# plt.plot(history.history['val_acc'])
# plt.title('model accuracy')
# plt.ylabel('accuracy')
# plt.xlabel('epoch')
# plt.legend(['train', 'test'], loc='upper left')
# plt.savefig(os.path.join('figures', CID + opts.title + 'learning-c.svg'),format='svg')
# plt.close()
#Y_de = decode_y(Y_test, features=enc.active_features_)
#Y_pred = model.predict(X_test)
#Y_depred = decode_y(Y_pred, features=enc.active_features_)
#print(classification_report(Y_de, Y_depred))
return