def get_train_data(perc, freq):
print("Feature selection percentile: %d" % perc)
train_data = load_train_data("./data/train.txt",
word_count_threshold=5,
freq=freq,
feature_selection_flag=True,
percentile=perc)
return train_data
def test_bnbc():
print_title("Binomial NBC")
#percentile_list = [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
percentile_list = [40]
for perc in percentile_list:
print("=" * 20)
print("Feature selection percentile: %d" % perc)
train_data = load_train_data(
"./data/train.txt", word_count_threshold=2, freq=False,
feature_selection_flag=True, percentile=perc)
bnbc_func(train_data["x"], train_data["y"])
def test_entropy():
print_title("Max-Entropy Classifier")
percentile_list = [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
#percentile_list = [40]
for perc in percentile_list:
print("=" * 20)
print("Feature selection percentile: %d" % perc)
train_data = load_train_data(
"./data/train.txt", word_count_threshold=1, freq=False,
feature_selection_flag=True, percentile=perc)
entropy_func(train_data['x'], train_data['y'])
def test_svm(kernel):
freq = False
print_title("SVM Classifier")
percentile_list = [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
#percentile_list = [40]
for perc in percentile_list:
print("=" * 20 + "\nkernel: %s || freq: %s" % (kernel, freq))
print("Feature selection percentile: %d" % perc)
train_data = load_train_data(
"./data/train.txt", word_count_threshold=1, freq=freq,
feature_selection_flag=True, percentile=perc)
svm_func(train_data['x'], train_data['y'], kernel=kernel)
