组合特征（四）linearsvm-tfidf(word)+lr-tfidf(article)

最新推荐文章于 2024-03-08 16:29:57 发布

Datawhale

最新推荐文章于 2024-03-08 16:29:57 发布

阅读量1k

点赞数

CC 4.0 BY-SA版权

分类专栏：达观杯nlp算法比赛总结

本文链接：https://blog.youkuaiyun.com/Datawhale/article/details/82911839

达观杯nlp算法比赛总结专栏收录该内容

20 篇文章

订阅专栏

本文介绍了一种结合线性SVM挑选的TF-IDF(word)特征与LR挑选的TF-IDF(article)特征的方法，通过使用pickle模块加载两个不同模型选择的特征集，并利用scipy.sparse.hstack函数将它们水平堆叠，最终合并成新的训练和测试数据集。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

"""
将linearsvm挑选的tfidf(word)特征和lr挑选的tfidf(article)

"""

import pickle
from scipy import sparse
from scipy.sparse import hstack

with open('linearsvm-tfidf(word).pkl', 'rb')as f_1:
    x_train_1, y_train, x_test_1 = pickle.load(f_1)


with open('lr-tfidf(article).pkl', 'rb')as f_2:
    x_train_2, y_train, x_test_2 = pickle.load(f_2)
    

x_train = hstack([x_train_1, x_train_2])
x_test = hstack([x_test_1, x_test_2])

data = (x_train, y_train, x_test)
with open('linearsvm-tfidf(word)+lr-tfidf(article).pkl', 'wb')as f:
    pickle.dump(data, f)