float(line.rstrip('\n')) for line in open('csvData/test_label.data')
]
filename = 'ensembleResult/test' + '_ensemble_SVM_Perceptron'
predictTest = [
1 if s == 1 and p == 1 else -1
for p, s in zip(predict_labels_Perceptron, predict_labels_SVM)
]
predictEval = [
1 if s == 1 and p == 1 else -1
for p, s in zip(eval_labels_Perceptron, eval_labels_SVM)
]
print("Accuracy for test set")
print(Stat.Accuracy(predictTest, test_labels))
print("F1 score for test set")
print(Stat.F1_Score(predictTest, test_labels))
print("Precision for test set")
print(Stat.Precision(predictTest, test_labels))
print("Recall for test set")
print(Stat.Recall(predictTest, test_labels))
# filename = 'results/eval_Ensemble_SVM_Perceptron'
#
# with open(filename, 'wb') as thefile:
# for item in predictEval:
# thefile.write("%s\n" % item)
print("best gamma0:", best_g0)
'''
###
sgd = SGD_SVM.SGDSVM(C=best_C,
ro=best_ro,
epochs=best_epoch,
W0=[0] * len(data[0]),
gamma0=best_g0)
sgd.fit(data, data_labels)
predictTrain = sgd.predict(data)
predictTest = sgd.predict(testset)
print("Accuracy for training set:")
print(Stat.Accuracy(predictTrain, data_labels))
print("F1 score for training set:")
print(Stat.F1_Score(predictTrain, data_labels))
print("Precision for training set:")
print(Stat.Precision(predictTrain, data_labels))
print("Recall for training set:")
print(Stat.Recall(predictTrain, data_labels))
print("Accuracy for test set")
print(Stat.Accuracy(predictTest, test_labels))
print("F1 score for test set")
print(Stat.F1_Score(predictTest, test_labels))
for epoch in [20]:#,10,15,25]:
for g0 in [1.1,1.01,1.001]:
tmp = []
sgd = SGD_SVM.SGDSVM(C=C,ro=ro,epochs=epoch,W0=[0]*len(phiTrain[0]),gamma0=g0)
kfold = KFold.KFold(n_splits=5)
for kf in kfold.split(phiTrain):
train2 = [phiTrain[i] for i in kf[0]]
train_label2 = [data_labels[i] for i in kf[0]]
test2 = [phiTrain[i] for i in kf[1]]
test_label2 = [data_labels[i] for i in kf[1]]
sgd.fit(train2, train_label2)
predict_tmp = sgd.predict(test2)
tmp.append(Stat.Accuracy(predict_tmp,test_label2))
if np.mean(tmp) > best_accuracy:
best_accuracy = np.mean(tmp)
best_C = C
best_epoch = epoch
best_g0 = g0
best_ro = ro
###
print("mid cross-validation")
sgd = SGD_SVM.SGDSVM(C=best_C,ro=best_ro,epochs=best_epoch,W0=[0]*len(phiTrain[0]),gamma0=best_g0)
sgd.fit(phiTrain,train_label)