def train_test():
X_features, y = create_training_vectors()
n_folds = 10
kf = StratifiedKFold(n_splits=n_folds, shuffle=True)
kf.get_n_splits(X_features, y)
total_acc, total_pre, total_recall, total_macro_f1, total_micro_f1 = [], [], [], [], []
for train_index, test_index in kf.split(X_features, y):
print("TRAIN:", train_index, "TEST:", test_index)
X_train, X_test = X_features[train_index], X_features[test_index]
y_train, y_test = y[train_index], y[test_index]
acc, pre, recall, macro_f1, micro_f1 = classify(train_X=X_train,
train_y=y_train,
test_X=X_test,
test_y=y_test)
total_acc.append(acc)
total_pre.append(pre)
total_recall.append(recall)
total_macro_f1.append(macro_f1)
total_micro_f1.append(micro_f1)
del X_train, X_test, y_train, y_test
print("======================")
print("avg acc:", np.mean(total_acc))
print("avg pre:", np.mean(total_pre))
print("avg recall:", np.mean(total_recall))
print("avg macro_f1:", np.mean(total_macro_f1))
print("avg micro_f1:", np.mean(total_micro_f1))
print("======================")
def train_test():
contents, y = create_training_content()
tfidf = MYTFIDF()
tfidf.tfidf_model_path = "/media/iiip/Elements/数据集/user_profiling/weibo/cache/tfidf_model.m"
X_features = tfidf.get_tfidf_vector(corpus=contents)
print(np.shape(X_features))
del contents
n_folds = 10
kf = StratifiedKFold(n_splits=n_folds, shuffle=True)
kf.get_n_splits(X_features, y)
total_acc, total_pre, total_recall, total_macro_f1, total_micro_f1 = [], [], [], [], []
for train_index, test_index in kf.split(X_features, y):
print("TRAIN:", train_index, "TEST:", test_index)
X_train, X_test = X_features[train_index], X_features[test_index]
y_train, y_test = y[train_index], y[test_index]
acc, pre, recall, macro_f1, micro_f1 = classify(train_X=X_train,
train_y=y_train,
test_X=X_test,
test_y=y_test)
total_acc.append(acc)
total_pre.append(pre)
total_recall.append(recall)
total_macro_f1.append(macro_f1)
total_micro_f1.append(micro_f1)
del X_train, X_test, y_train, y_test
print("======================")
print("avg acc:", np.mean(total_acc))
print("avg pre:", np.mean(total_pre))
print("avg recall:", np.mean(total_recall))
print("avg macro_f1:", np.mean(total_macro_f1))
print("avg micro_f1:", np.mean(total_micro_f1))
print("======================")
def train_chi2_model():
contents, y = create_training_content()
contents = np.array(contents)
chi2 = MYCHI2()
chi2.cv_model_path = "/media/iiip/Elements/数据集/user_profiling/weibo/cache/cv_model.m"
chi2.chi2_model_path = "/media/iiip/Elements/数据集/user_profiling/weibo/cache/chi2_model.m"
n_folds = 10
kf = StratifiedKFold(n_splits=n_folds, shuffle=True)
kf.get_n_splits(contents, y)
total_acc, total_pre, total_recall, total_macro_f1, total_micro_f1 = [], [], [], [], []
for train_index, test_index in kf.split(contents, y):
print("TRAIN:", train_index, "TEST:", test_index)
X_train, X_test = contents[train_index], contents[test_index]
y_train, y_test = y[train_index], y[test_index]
chi2.train_chi2(X_train, y_train)
X_chi2_train = chi2.get_chi2_vector(X_train)
X_chi2_test = chi2.get_chi2_vector(X_test)
del X_train, X_test
acc, pre, recall, macro_f1, micro_f1 = classify(train_X=X_chi2_train, train_y=y_train, test_X=X_chi2_test, test_y=y_test)
total_acc.append(acc)
total_pre.append(pre)
total_recall.append(recall)
total_macro_f1.append(macro_f1)
total_micro_f1.append(micro_f1)
del X_chi2_train, X_chi2_test, y_train, y_test
print("======================")
print("avg acc:", np.mean(total_acc))
print("avg pre:", np.mean(total_pre))
print("avg recall:", np.mean(total_recall))
print("avg macro_f1:", np.mean(total_macro_f1))
print("avg micro_f1:", np.mean(total_micro_f1))
print("======================")