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
- c语言的编译过程和GCC 编译参数
原文: http://www.cnblogs.com/zhangShanGui/p/4912135.html C语言的编译过程和GCC编译参数 C语言的编译一般有三个步骤: 预编译: gcc -E - ...
- CSS中如何把Span标签设置为固定宽度
一.形如<span>ABC</span>独立行设置SPAN为固定宽度方法如下: span {width:60px; text-align:center; display:blo ...
- tab奇偶行颜色交替+插件
(function($){ $.fn.tableUI=function(options){ var defaults={ evenRowclass:"evenRow", oddro ...
- About struct in C
something new: to set value in struct can be in case i cannot view picture.. i write the snippet her ...
- robot framework -记录错误
1.注意ie浏览器代理设置,报奇怪的错误 2.注意浏览器的安全设置,保护模式全部不要勾选
- PL/SQL Developer StringBuffer 专用复制
;PL/SQL Developer SpecialCopy definition;<LINE_1> for first line;<LINE_*> for all other ...
- angularjs model.service vs provider vs factory?
<!DOCTYPE html> <html ng-app="app"> <head> <script src="http://c ...
- linux 命令实现原理
我们知道有些Linux的命令涉及到一些高效率的算法,在此做出一个积累吧,不是系统的. 1.tail命令打印一个文件的最后num行 2.grep命令从文本中匹配字符串 基于正则表达式的匹配很快. it ...
- leetcode202（Floyd判圈算法（龟兔赛跑算法））
Write an algorithm to determine if a number is "happy". 写出一个算法确定一个数是不是快乐数. A happy number ...
- webapp之路--meta标签format-detection、apple-mobile-web-app-capable
1. format-detection翻译成中文的意思是“格式检测”,顾名思义,它是用来检测html里的一些格式的,那关于meta的format-detection属性主要是有以下几个设置: meta ...