Python数据挖掘入门与实践 数据集及代码

Python数据挖掘入门与实践 数据集及代码

import os
import re
from mrjob.job import MRJob
from mrjob.step import MRStep

word_search_re = re.compile（r“[\w‘]+“）


class ExtractPosts（MRJob）:

    post_start = False
    post = []

    def mapper（self key line）:
        filename = os.environ[“map_input_file“]
        gender = filename.split（“.“）[1]
        try:
            docnum = int（filename[0]）
        except:
            docnum = 8
        if filename.startswith（“51“）:
            # remove leading and trailing whitespace
            line = line.strip（）
            if line == ““:
                self.post_start = True
            elif line == ““:
                self.post_start = False
                yield gender repr（“\n“.join（self.post））
                self.post = []
            elif self.post_start:
                self.post.append（line）



if __name__ == ‘__main__‘:
    ExtractPosts.run（）

 属性            大小     日期    时间   名称
----------- ---------  ---------- -----  ----
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\
     文件         764  2016-09-27 07:16  LearningDataMiningWithPython-master\.gitignore
     文件        1112  2016-09-27 07:16  LearningDataMiningWithPython-master\INSTALL.md
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 1\
     文件        1000  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 1\affinity_dataset.txt
     文件       16777  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 1\ch1_affinity.ipynb
     文件        3670  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 1\ch1_affinity_create.ipynb
     文件       13847  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 1\ch1_oner_application.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 10\
     文件       78925  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 10\Chapter 10 Clusterer.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 11\
     文件       59206  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 11\Chapter 11 （CIFAR）.ipynb
     文件       62409  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 11\Chapter 11 （Theano and Lasagne）.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\
     文件       38759  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\CH12 MapReduce Basics.ipynb
     文件       10578  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\Chapter 12 （NB Predict）.ipynb
     文件        1730  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\Chapter 12 （Test load）.ipynb
     文件         882  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\extract_posts.py
     文件        1986  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\nb_predict.py
     文件        2021  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 12\nb_train.py
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 2\
     文件      143291  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 2\Ionosphere Nearest Neighbour.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 3\
     文件       45385  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 3\Basketball Results.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 4\
     文件       46259  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 4\ch4 Affinity Analysis.ipynb
     目录           0  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 5\
     文件        1034  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 5\adult_tests.py
     文件       13293  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 5\ch5_adult.ipynb
     文件       29985  2016-09-27 07:16  LearningDataMiningWithPython-master\LearningDataMiningBook\Chapter 5\ch5_advertisements.ipynb
............此处省略18个文件信息