def evaluate(tokenizer, terror_cat, texts, cats):
tokens_list = (tokenizer(text) for text in texts)
labels, scores = list(), list()
for i, doc in enumerate(terror_cat.pipe(tokens_list)):
label = cats[i]['cats'][true_label]
score = doc.cats[true_label]
labels.append(label)
scores.append(score)
au.precision_recall_threshold(labels, scores)
print('\n')
def test(self, test_file):
textarr, labelarr = file2label_text_array(test_file)
featurearr = self.textarr2featurearr(textarr)
probarr = self.predict_proba(featurearr)
au.precision_recall_threshold(
labelarr,
probarr,
file="performance.csv",
thres_range=[i / 100 for i in range(1, 10)] +
[i / 20 for i in range(2, 20)])
fu.dump_array("result.json", (labelarr, probarr))
def test(self, test_file):
textarr, labelarr = file2label_text_array(test_file)
""""""
# docarr = su.textarr_nlp(textarr, self.get_nlp())
# featurearr = self.textarr2featurearr(textarr, docarr)
featurearr = self.textarr2featurearr_no_gpe(textarr)
""""""
probarr = self.predict_proba(featurearr)
au.precision_recall_threshold(labelarr, probarr,
thres_range=[i / 100 for i in range(1, 10)] + [i / 20 for i in range(2, 20)])
fu.dump_array("result.json", (labelarr, probarr))
def test(self, test_file):
"""
给定带标记和文本的文件，读取其中的文本-标记对，调用向量化接口以及分类器，评估分类器在测试集上的性能
:param test_file: str，测试用文本文件的路径
:return:
"""
textarr, labelarr = file2label_text_array(test_file)
""""""
# docarr = su.textarr_nlp(textarr, self.get_nlp())
# featurearr = self.textarr2featurearr(textarr, docarr)
featurearr = self.textarr2featurearr_no_gpe(textarr)
""""""
probarr = self.predict_proba(featurearr)
au.precision_recall_threshold(
labelarr,
probarr,
thres_range=[i / 100 for i in range(1, 10)] +
[i / 20 for i in range(2, 20)])
fu.dump_array("result.json", (labelarr, probarr))
def train_keyword_similarity(exec_train=True):
np.random.seed(134590)
data = np.load('./data/data_sim.npy')
data_len = len(data)
sep_idx = int(data_len * 0.8)
# train, test = data[:sep_idx], data[sep_idx:]
rand_idx = au.shuffle([i for i in range(data_len)])
train, test = data[rand_idx[:sep_idx]], data[rand_idx[sep_idx:]]
train_x, train_y = train[:, :-1], train[:, -1]
test_x, test_y = test[:, :-1], test[:, -1]
print(data.shape, train.shape, test.shape)
if exec_train:
clf = LogisticRegression()
clf.fit(train_x, train_y)
joblib.dump(clf, './data/judge_merge_model')
else:
clf = joblib.load('./data/judge_merge_model')
# print("coef:", c.coef_.tolist())
pred = clf.predict_proba(test_x)[:, 1]
au.precision_recall_threshold(test_y, pred, thres_range=[i / 10 for i in range(1, 10)])
def perfomance_analysis():
labal, proba = fu.load_array('label_proba')
print(len(labal), len(proba))
au.precision_recall_threshold(labal, proba)
# post_twarr = list()
# for idx in range(len(probarr)):
# if probarr[idx] >= 0.35:
# post_twarr.append(twarr[idx])
# else:
# print(twarr[idx][tk.key_text])
# post_twarr = [tw for idx, tw in enumerate(twarr) if probarr[idx] >= 0.4]
# post_total_len += len(post_twarr)
# print(len(post_twarr) / len(twarr), '\n\n\n')
tmu.check_time()
lblarr = [1 for _ in range(len(pos_probarr))] + [0 for _ in range(len(neg_probarr))]
prbarr = pos_probarr + neg_probarr
fu.dump_array("prb_lbl_arr.txt", (lblarr, prbarr))
lblarr, prbarr = fu.load_array("prb_lbl_arr.txt")
au.precision_recall_threshold(lblarr, prbarr)
# print('total portion = {} / {} = {}'.format(post_total_len, pre_total_len, post_total_len / pre_total_len))
tmu.check_time()
exit()
sub_files = fi.listchildren('/home/nfs/cdong/tw/origin/', fi.TYPE_FILE, concat=True)[18:19]
twarr = au.merge_array([fu.load_array(file) for file in sub_files])
print(len(twarr))
tmu.check_time(print_func=None)
for idx, tw in enumerate(twarr[14000:15000]):
if (idx + 1) % 1000 == 0:
print(idx)
try:
my_filter.get_features(tw)
except:
# print(tw[tk.key_text])
import matplotlib.pyplot as plt
import utils.function_utils as fu
def figure(X, Y, fig_name):
# plt.figure(figsize=(13, 7))
plt.plot(X, Y, color="blue", linewidth=1)
plt.xlim([-0.03, 1.03])
plt.ylim([-0.03, 1.03])
plt.ylabel('True Positive Rate')
plt.xlabel('False Positive Rate')
plt.title("roc curve")
plt.legend(loc='lower right')
plt.savefig(fig_name, format='png')
if __name__ == '__main__':
import utils.array_utils as au
labelarr, probarr = fu.load_array(
"/home/nfs/cdong/tw/src/preprocess/filter/prb_lbl_arr.txt")
from sklearn.metrics import roc_curve
fpr, tpr, thresholds = roc_curve(labelarr, probarr)
figure(fpr, tpr, "/home/nfs/cdong/tw/src/preprocess/filter/roc_curve.png")
au.precision_recall_threshold(
labelarr,
probarr,
file="/home/nfs/cdong/tw/src/preprocess/filter/performance.csv",
thres_range=[i / 100
for i in range(1, 10)] + [i / 20 for i in range(2, 20)])
