Classification of text documents: using a MLComp dataset
注：原文代码链接http://scikit-learn.org/stable/auto_examples/text/mlcomp_sparse_document_classification.html
运行结果为：
Loading 20 newsgroups training set...
20 newsgroups dataset for document classification (http://people.csail.mit.edu/jrennie/20Newsgroups)
13180 documents
20 categories
Extracting features from the dataset using a sparse vectorizer
done in 139.231000s
n_samples: 13180, n_features: 130274
Loading 20 newsgroups test set...
done in 0.000000s
Predicting the labels of the test set...
5648 documents
20 categories
Extracting features from the dataset using the same vectorizer
done in 7.082000s
n_samples: 5648, n_features: 130274
Testbenching a linear classifier...
parameters: {'penalty': 'l2', 'loss': 'hinge', 'alpha': 1e-05, 'fit_intercept': True, 'n_iter': 50}
done in 22.012000s
Percentage of non zeros coef: 30.074190
Predicting the outcomes of the testing set
done in 0.172000s
Classification report on test set for classifier:
SGDClassifier(alpha=1e-05, average=False, class_weight=None, epsilon=0.1,
eta0=0.0, fit_intercept=True, l1_ratio=0.15,
learning_rate='optimal', loss='hinge', n_iter=50, n_jobs=1,
penalty='l2', power_t=0.5, random_state=None, shuffle=True,
verbose=0, warm_start=False) precision recall f1-score support alt.atheism 0.95 0.93 0.94 245
comp.graphics 0.85 0.91 0.88 298
comp.os.ms-windows.misc 0.88 0.88 0.88 292
comp.sys.ibm.pc.hardware 0.82 0.80 0.81 301
comp.sys.mac.hardware 0.90 0.92 0.91 256
comp.windows.x 0.92 0.88 0.90 297
misc.forsale 0.87 0.89 0.88 290
rec.autos 0.93 0.94 0.94 324
rec.motorcycles 0.97 0.97 0.97 294
rec.sport.baseball 0.97 0.97 0.97 315
rec.sport.hockey 0.98 0.99 0.99 302
sci.crypt 0.97 0.96 0.96 297
sci.electronics 0.87 0.89 0.88 313
sci.med 0.97 0.97 0.97 277
sci.space 0.97 0.97 0.97 305
soc.religion.christian 0.95 0.96 0.95 293
talk.politics.guns 0.94 0.94 0.94 246
talk.politics.mideast 0.97 0.99 0.98 296
talk.politics.misc 0.96 0.92 0.94 236
talk.religion.misc 0.89 0.84 0.86 171 avg / total 0.93 0.93 0.93 5648 Confusion matrix:
[[227 0 0 0 0 0 0 0 0 0 0 1 2 1 1 1 0 1
0 11]
[ 0 271 3 8 2 5 2 0 0 1 0 0 3 1 1 0 0 1
0 0]
[ 0 7 256 14 5 6 1 0 0 0 0 0 2 0 1 0 0 0
0 0]
[ 1 8 12 240 9 3 12 2 0 0 0 1 12 0 0 1 0 0
0 0]
[ 0 1 3 6 235 2 4 0 0 0 0 1 3 0 1 0 0 0
0 0]
[ 0 17 9 4 0 260 0 0 1 1 0 0 2 0 2 0 1 0
0 0]
[ 0 1 3 7 3 0 257 7 2 0 0 1 8 0 1 0 0 0
0 0]
[ 0 0 0 2 1 0 5 305 2 3 0 0 4 1 0 0 1 0
0 0]
[ 0 0 0 0 1 0 3 3 285 0 0 0 1 0 0 1 0 0
0 0]
[ 0 0 0 0 0 0 3 2 0 305 2 1 1 0 0 0 0 0
1 0]
[ 0 0 0 0 0 0 1 0 1 0 300 0 0 0 0 0 0 0
0 0]
[ 0 0 1 1 0 2 0 1 0 0 0 284 0 1 1 0 2 2
1 1]
[ 0 2 2 10 2 2 6 5 1 0 1 1 279 1 1 0 0 0
0 0]
[ 0 3 0 0 1 1 1 0 0 0 0 0 0 269 0 1 1 0
0 0]
[ 0 5 0 0 1 0 0 0 0 0 2 0 1 0 295 0 0 0
1 0]
[ 1 1 1 0 0 1 0 1 0 0 0 0 0 1 1 282 1 0
0 3]
[ 0 0 1 0 0 0 0 0 1 3 0 0 1 0 0 1 232 1
5 1]
[ 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 2 0 293
0 0]
[ 0 2 0 0 0 0 2 0 0 1 0 1 0 1 0 0 7 4
216 2]
[ 11 0 0 0 0 0 0 0 0 0 0 1 0 2 0 9 2 1
2 143]]
Testbenching a MultinomialNB classifier...
parameters: {'alpha': 0.01}
done in 0.608000s
Percentage of non zeros coef: 100.000000
Predicting the outcomes of the testing set
done in 0.203000s
Classification report on test set for classifier:
MultinomialNB(alpha=0.01, class_prior=None, fit_prior=True) precision recall f1-score support alt.atheism 0.90 0.92 0.91 245
comp.graphics 0.81 0.89 0.85 298
comp.os.ms-windows.misc 0.87 0.83 0.85 292
comp.sys.ibm.pc.hardware 0.82 0.83 0.83 301
comp.sys.mac.hardware 0.90 0.92 0.91 256
comp.windows.x 0.90 0.89 0.89 297
misc.forsale 0.90 0.84 0.87 290
rec.autos 0.93 0.94 0.93 324
rec.motorcycles 0.98 0.97 0.97 294
rec.sport.baseball 0.97 0.97 0.97 315
rec.sport.hockey 0.97 0.99 0.98 302
sci.crypt 0.95 0.95 0.95 297
sci.electronics 0.90 0.86 0.88 313
sci.med 0.97 0.96 0.97 277
sci.space 0.95 0.97 0.96 305
soc.religion.christian 0.91 0.97 0.94 293
talk.politics.guns 0.89 0.96 0.93 246
talk.politics.mideast 0.95 0.98 0.97 296
talk.politics.misc 0.93 0.87 0.90 236
talk.religion.misc 0.92 0.74 0.82 171 avg / total 0.92 0.92 0.92 5648 Confusion matrix:
[[226 0 0 0 0 0 0 0 0 1 0 0 0 0 2 7 0 0
0 9]
[ 1 266 7 4 1 6 2 2 0 0 0 3 4 1 1 0 0 0
0 0]
[ 0 11 243 22 4 7 1 0 0 0 0 1 2 0 0 0 0 0
1 0]
[ 0 7 12 250 8 4 9 0 0 1 1 0 9 0 0 0 0 0
0 0]
[ 0 3 3 5 235 2 3 1 0 0 0 2 1 0 1 0 0 0
0 0]
[ 0 19 5 3 2 263 0 0 0 0 0 1 0 1 1 0 2 0
0 0]
[ 0 1 4 9 3 1 243 9 2 3 1 0 8 0 0 0 2 2
2 0]
[ 0 0 0 1 1 0 5 304 1 2 0 0 3 2 3 1 1 0
0 0]
[ 0 0 0 0 0 2 2 3 285 0 0 0 1 0 0 0 0 0
0 1]
[ 0 1 0 0 0 1 1 3 0 304 5 0 0 0 0 0 0 0
0 0]
[ 0 0 0 0 0 0 0 0 1 2 299 0 0 0 0 0 0 0
0 0]
[ 0 2 2 1 0 1 2 0 0 0 0 283 1 0 0 0 2 1
2 0]
[ 0 11 1 9 3 1 3 5 1 0 1 4 270 1 3 0 0 0
0 0]
[ 0 2 0 1 1 1 0 0 0 0 0 1 0 266 2 1 0 0
2 0]
[ 0 2 0 0 1 0 0 0 0 0 0 2 1 1 296 0 1 1
0 0]
[ 3 1 0 0 0 0 0 0 0 0 1 0 0 2 0 283 0 1
2 0]
[ 1 0 1 0 0 0 0 0 1 0 0 1 0 0 0 0 237 1
3 1]
[ 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 3 0 291
0 0]
[ 1 1 0 0 1 1 0 1 0 0 0 0 0 0 1 1 17 6
206 0]
[ 18 1 0 0 0 0 0 0 0 1 0 0 0 0 0 14 4 2
4 127]]
步骤为：
一、preprocessing
1.加载训练集（training set）
2.训练集特征提取，用TfidfVectorizer，得到训练集上的x_train和y_train
3.加载测试集（test set）
4.测试集特征提取，用TfidfVectorizer，得到测试集上的x_train和y_train
二、定义Benchmark classifiers
5.训练，clf = clf_class(**params).fit(X_train, y_train)
6.测试，pred = clf.predict(X_test)
7.测试集上分类报告，print(classification_report(y_test, pred,target_names=news_test.target_names))
8.confusion matrix，cm = confusion_matrix(y_test, pred)
三、训练
9.调用两个分类器，SGDClassifier和MultinomialNB
Classification of text documents: using a MLComp dataset的更多相关文章
- Clustering text documents using k-means
源代码的链接为http://scikit-learn.org/stable/auto_examples/text/document_clustering.html Loading 20 newsgro ...
- scikit-learn：4.2.3. Text feature extraction
http://scikit-learn.org/stable/modules/feature_extraction.html 4.2节内容太多,因此将文本特征提取单独作为一块. 1.the bag o ...
- Python scikit-learn机器学习工具包学习笔记
feature_selection模块 Univariate feature selection:单变量的特征选择 单变量特征选择的原理是分别单独的计算每个变量的某个统计指标,根据该指标来判断哪些指标 ...
- 特征选择 (feature_selection)
目录 特征选择 (feature_selection) Filter 1. 移除低方差的特征 (Removing features with low variance) 2. 单变量特征选择 (Uni ...
- sklearn—特征工程
sklearn实战-乳腺癌细胞数据挖掘(博主亲自录制视频) https://study.163.com/course/introduction.htm?courseId=1005269003& ...
- scikit-learn：3.3. Model evaluation: quantifying the quality of predictions
參考:http://scikit-learn.org/stable/modules/model_evaluation.html#scoring-parameter 三种方法评估模型的预測质量: Est ...
- [Scikit-learn] 1.1 Generalized Linear Models - Comparing various online solvers
数据集分割 一.Online learning for 手写识别 From: Comparing various online solvers An example showing how diffe ...
- [Scikit-learn] Yield miniBatch for online learning.
From: Out-of-core classification of text documents Code: """ ======================= ...
- sklearn中的模型评估-构建评估函数
1.介绍 有三种不同的方法来评估一个模型的预测质量: estimator的score方法:sklearn中的estimator都具有一个score方法,它提供了一个缺省的评估法则来解决问题. Scor ...
随机推荐
- Hibernate 系列教程17-查询缓存
在二级缓存配置成功的基础上进行查询缓存配置 Product public class Product { private Long id; private String name; Product.h ...
- phpStudy环境配置多个站点，绑定域名
经常做网站的朋友,往往要在自已的电脑同时建立多个站点,而phpstudy这款软件就能很好的解决这个问题,大家看下图 点击上图中的 其它选项菜单 ,就会弹出下面的对话框,然后点击 站点域名管理 然 后在 ...
- 关于tag,viewWithTag
iOS SDK内置了一套搜寻机制,可通过tag来查找子视图. **苹果公司很少给子视图设置tag.笔者所知范围的唯一例外出现在UIAlertView中,该类会给按钮分别设置值为1.2的标签 viewW ...
- Linux 安全模块
LSM是Linux Secrity Module的简称,即linux安全模块.其是一种轻量级通用访问控制框架,适合于多种访问控制模型在它上面以内核可加载模块的形实现.用户可以根据自己的需求选择合适的安 ...
- 求n!末尾0的个数
题目连接 /* £:离散数学. £:n!中2的个数>5的个数. £:2*5=10: */ #include<cstdio> #include<cstring> #incl ...
- J2SE基本数据结构
1.J2SE中的常用数据结构对象的继承关系如下图 Collection ........|--------List ........|..........|----------ArrayList .. ...
- loadrunner多负载机联合产生负载
解说一: 多机联合产生负载 LOADRUNNER 对应用程序施压时,采用的方法就是让一台机器模拟很多用户,同时向被 测用户发送请求或进行操作.这样,如果一台测试机器模拟的虚拟用户数过多,他本身性能的下 ...
- scp命令报错-bash: scp: command not found
# scp -bash: scp: command not found # which scp /usr/bin/scp # rpm -qf /usr/bin/scp openssh-clients- ...
- Cells Not Under Attack
Cells Not Under Attack Vasya has the square chessboard of size n × n and m rooks. Initially the ches ...
- UITabBar 蓝色
效果图1: 第一种解决办法(有局限性): 但是,但是!!!!!虽然不用写代码看起来好方便,在iOS9和8上貌似都没问题.然后我默默地 拿出了自己的小4,发现还似蓝色... 第二种解决办法:(彻底): ...