def nn_feature_selection_wrap(training_feature,training_label,alpha):
Result_List = CalParList (3,"alpha","feature_0","feature_1")
skf = StratifiedKFold(n_splits=10,shuffle=True)
skf.get_n_splits(training_feature,training_label['label'])
feature_avaliable = ['feature0','feature1','feature2','feature3','feature4']
feature_choice=list(itertools.combinations(feature_avaliable ,2))
for i in range(len(feature_choice)):
Cal_Result_List = CalList ()
for train_index, test_index in skf.split(training_feature,training_label['label']):
X_train, X_val = training_feature.loc[train_index], training_feature.loc[test_index]
y_train, y_val = training_label.loc[train_index], training_label.loc[test_index]
X_train=X_train.loc[:,[feature_choice[i][0],feature_choice[i][1]]]
X_val=X_val.loc[:,[feature_choice[i][0],feature_choice[i][1]]]
dis_1,dis_2=nn_distance_calculate(X_val,X_train,y_train)
y_pred_temp=nn_predict(dis_1,dis_2,alpha,X_val)
Precall,f1_score,BER,FPR = cal_score (y_pred_temp,y_val['label'])
Cal_Result_List.list_append(Precall,f1_score,BER,FPR)
Precall,FPR,BER,f1_score = Cal_Result_List.list_average_cal()
Result_List.list_append (Precall,f1_score,BER,FPR,alpha,feature_choice[i][0],feature_choice[i][1])
result = Result_List.return_result()
return result
def parameter_adjust(X_train, y_train, sample_amount, data_ratio):
start = time.time()
label_1_amount = int(sample_amount * (data_ratio / (data_ratio + 1)))
label_2_amount = int(sample_amount - label_1_amount)
Result_List = CalParList(3, "gamma_exp", "C_exp", "time")
for gamma_exp in [-15, -13, -11, -9, -7, -5, -3, -1, 1, 3]:
for C_exp in [-5, -3, -1, 1, 3, 5, 7, 9, 11, 13, 15]:
start1 = time.time()
sample_feature, sample_label = choose_data_seperately(
X_train, y_train, label_1_amount, label_2_amount)
tpr, fpr, BER, f1_score, time_var = SVM_cross_validation(
sample_feature, sample_label, 2**C_exp, 2**gamma_exp)
Result_List.list_append(tpr, f1_score, BER, fpr, gamma_exp, C_exp,
time_var)
#print("fit time:%5.1fminute"%(temp))
print("the total executing time:%5.1fminute" %
((time.time() - start) / 60))
result = Result_List.return_result()
return result
def nn_predict_with_distance_adjust_presion (training_feature,training_label,alpha_lower_bound,alpha_higher_bound):
#dis_1,dis_2=nn_distance_calculate(X_val,X_train,y_train)
alpha =alpha_lower_bound
Result_List = CalParList (1,"alpha")
while(alpha<=alpha_higher_bound):
Precall,FPR,BER,f1_score = NN_cross_validation(training_feature,training_label,alpha)
Result_List.list_append (Precall,f1_score,BER,FPR,alpha)
alpha=alpha+0.001
result = Result_List.return_result()
return result
def kNN_data_ratio_adjust(training_feature, training_label, k_value):
start = time.time()
label_1_amount = 40000
label_2_amount = 10000
Result_List = CalParList(4, "label_1_amount", "label_2_amount", "ratio",
"time")
iter_amount = 5
train_data = pd.concat([training_feature, training_label['label']],
axis=1,
join='outer')
while (label_1_amount > 2000):
count = 0
Cal_Result_List = CalList()
time_list_temp = []
while (count < iter_amount):
start1 = time.time()
sample_feature, sample_label = choose_data_seperately(
training_feature, training_label, label_1_amount,
label_2_amount)
Precall, FPR, BER, f1_score, time_var = kNN_cross_validation(
sample_feature, sample_label, k_value)
Cal_Result_List.list_append(Precall, f1_score, BER, FPR)
time_list_temp.append(time_var)
count = count + 1
Precall, FPR, BER, f1_score = Cal_Result_List.list_average_cal()
time_ave = sum(time_list_temp) / len(time_list_temp)
Result_List.list_append(Precall, f1_score, BER, FPR, label_1_amount,
label_2_amount,
label_1_amount / label_2_amount, time_ave)
#print("current data labe 1 size:%d ,fit time:%5.1fminute"%(t,(time.time()-start1)/60))
if (label_1_amount > 10000):
label_1_amount = label_1_amount - 5000
else:
label_1_amount = label_1_amount - 2500
print("the total executing time:%5.1fminute" %
((time.time() - start) / 60))
result = Result_List.return_result()
return result
def nn_predict_with_distance_adjust (training_feature,training_label):#,#X_val,y_val):
Result_List = CalParList (1,"alpha")
alpha = 0.10
while(alpha<=0.9):
Precall,FPR,BER,f1_score = NN_cross_validation(training_feature,training_label,alpha)
Result_List.list_append (Precall,f1_score,BER,FPR,alpha)
if (0.4<=alpha<=0.6):
alpha=alpha+0.01
else:
alpha=alpha+0.05
result = Result_List.return_result()
return result
def sample_amount_choice(X_train, y_train, sample_amount_upper_bound):
start = time.time()
Result_List = CalParList(2, "sample_amount", "time")
sample_amount = sample_amount_upper_bound
iter_amount = 3
while (sample_amount > 800):
count = 0
Cal_Result_List = CalList()
time_list_temp = []
while (count < iter_amount):
print("current sample amount:%d" % sample_amount)
sample_feature, sample_label = choose_data_together(
X_train, y_train, sample_amount)
tpr, fpr, BER, f1_score, time_var = SVM_cross_validation(
sample_feature, sample_label, 1, 'auto')
Cal_Result_List.list_append(tpr, f1_score, BER, fpr)
time_list_temp.append(time_var)
count = count + 1
Precall, FPR, BER, f1_score = Cal_Result_List.list_average_cal()
time_ave = sum(time_list_temp) / len(time_list_temp)
Result_List.list_append(Precall, f1_score, BER, FPR, sample_amount,
time_ave)
if (sample_amount > 10000):
sample_amount = int(sample_amount / 2)
elif (sample_amount <= 2000):
sample_amount = sample_amount - 100
iter_amount = 10
else:
sample_amount = sample_amount - 2000
iter_amount = 10
print("the total executing time:%5.1fminute" %
((time.time() - start) / 60))
result = Result_List.return_result()
return result
def SVC_data_ratio_adjust(X_train, y_train, sample_amount):
start = time.time()
Result_List = CalParList(4, "label_1_amount", "label_2_amount", "ratio",
"time")
iter_amount = 5
data_ratio = 4
while (data_ratio > 0.2):
count = 0
Cal_Result_List = CalList()
time_list_temp = []
label_1_amount = int(sample_amount * (data_ratio / (data_ratio + 1)))
label_2_amount = int(sample_amount - label_1_amount)
while (count < iter_amount):
start1 = time.time()
sample_feature, sample_label = choose_data_seperately(
X_train, y_train, label_1_amount, label_2_amount)
tpr, fpr, BER, f1_score, time_var = SVM_cross_validation(
sample_feature, sample_label, 1, 'auto')
time_list_temp.append(time_var)
Cal_Result_List.list_append(tpr, f1_score, BER, fpr)
count = count + 1
Precall, FPR, BER, f1_score = Cal_Result_List.list_average_cal()
time_ave = sum(time_list_temp) / len(time_list_temp)
Result_List.list_append(Precall, f1_score, BER, FPR, label_1_amount,
label_2_amount,
label_1_amount / label_2_amount, time_ave)
if (data_ratio > 2):
data_ratio = data_ratio / 2
elif (data_ratio < 0.8):
data_ratio = data_ratio - 0.25
else:
data_ratio = data_ratio - 0.1
print("the total executing time:%5.1fminute" %
((time.time() - start) / 60))
result = Result_List.return_result()
return result
def kNN_k_parameter_adjust(training_feature, training_label):
#start=time.time()
k_value = 5
Result_List = CalParList(2, "k_value", "time")
while (k_value < 10):
Precall, FPR, BER, f1_score, time_var = kNN_cross_validation(
training_feature, training_label, k_value)
Result_List.list_append(Precall, f1_score, BER, FPR, k_value, time_var)
if (k_value < 27):
k_value = k_value + 2
else:
k_value = k_value + 5
#print("the total executing time:%5.1fminute"%((time.time()-start)/60))
result = Result_List.return_result()
return result