Adding comment.

Author: dashjim

ch03/noise_analysis.py

@@ -1,3 +1,4 @@
+# coding=utf-8
 # This code is supporting material for the book
 # Building Machine Learning Systems with Python
 # by Willi Richert and Luis Pedro Coelho
@@ -26,17 +27,18 @@ class StemmedTfidfVectorizer(TfidfVectorizer):
 
 defbuild_analyzer(self):
 analyzer=super(TfidfVectorizer, self).build_analyzer()
+# 下面代码似乎有问题？应该先Stem，再analyze
 returnlambdadoc: (english_stemmer.stem(w) forwinanalyzer(doc))
 
 vectorizer=StemmedTfidfVectorizer(min_df=10, max_df=0.5,
 stop_words='english', decode_error='ignore'
  )
-vectorized=vectorizer.fit_transform(train_data.data)
+vectorized=vectorizer.fit_transform(train_data.data)# TF-IDF 数字
 
 post_group=zip(train_data.data, train_data.target)
 # Create a list of tuples that can be sorted by
 # the length of the posts
-all= [(len(post[0]), post[0], train_data.target_names[post[1]])
+all= [(len(post[0]), post[0], train_data.target_names[post[1]])# post[1] 是数字，转换成名字
 forpostinpost_group]
 graphics=sorted([postforpostinallifpost[2] =='comp.graphics'])
 print(graphics[5])
@@ -46,13 +48,13 @@ def build_analyzer(self):
 # \n\n==============================================================================\n',
 # 'comp.graphics')
 
-noise_post=graphics[5][1]
+noise_post=graphics[5][1]# get the content -Jim
 
-analyzer=vectorizer.build_analyzer()
+analyzer=vectorizer.build_analyzer()# 第二个analyzer，确保还有stem功能？
 print(list(analyzer(noise_post)))
-
+# vectorizer.get_feature_names 返回全部4千多个词
 useful=set(analyzer(noise_post)).intersection(vectorizer.get_feature_names())
-print(sorted(useful))
+print(sorted(useful))# 去掉了近一般的词
 # ['ac', 'birmingham', 'host', 'kingdom', 'nntp', 'sorri', 'test', 'uk', 'unit', 'univers']
 
 forterminsorted(useful):

ch03/rel_post_01.py

@@ -1,3 +1,4 @@
+# coding=utf-8
 # This code is supporting material for the book
 # Building Machine Learning Systems with Python
 # by Willi Richert and Luis Pedro Coelho
@@ -51,25 +52,26 @@ def build_analyzer(self):
 print("#samples: %d, #features: %d"% (num_samples, num_features))
 
 new_post_vec=vectorizer.transform([new_post])
-print(new_post_vec, type(new_post_vec))
+print(new_post_vec, type(new_post_vec))# 稀疏表示
 print(new_post_vec.toarray())
 print(vectorizer.get_feature_names())
 
-
+# 不能处理重复的句子
 defdist_raw(v1, v2):
 delta=v1-v2
 returnsp.linalg.norm(delta.toarray())
 
-
+# 处理重复的句子 (什么鬼？两种方法结果一样？)
 defdist_norm(v1, v2):
 v1_normalized=v1/sp.linalg.norm(v1.toarray())
 v2_normalized=v2/sp.linalg.norm(v2.toarray())
 
 delta=v1_normalized-v2_normalized
-
+# norm()向量的长度
 returnsp.linalg.norm(delta.toarray())
 
 dist=dist_norm
+# dist = dist_raw (什么鬼？两种方法结果一样？)
 
 best_dist=sys.maxsize
 best_i=None

ch03/rel_post_20news.py

@@ -1,3 +1,4 @@
+# coding=utf-8
 # This code is supporting material for the book
 # Building Machine Learning Systems with Python
 # by Willi Richert and Luis Pedro Coelho
@@ -56,15 +57,18 @@ def build_analyzer(self):
  )
 
 vectorized=vectorizer.fit_transform(train_data.data)
+# shape 是一个tuple
 num_samples, num_features=vectorized.shape
 print("#samples: %d, #features: %d"% (num_samples, num_features))
 # samples: 3529, #features: 4712
 
 fromsklearn.clusterimportKMeans
-
+# make it verbose will print some information -jim
 km=KMeans(n_clusters=num_clusters, n_init=1, verbose=1, random_state=3)
+# 传给模型的是一个二维数组，一共有3529行，每一行代表一个post，每一行的宽度是4712，其中的数字是TF-IDF的得分
 clustered=km.fit(vectorized)
 
+# 模型结果的解读...
 print("km.labels_=%s"%km.labels_)
 # km.labels_=[ 6 34 22 ..., 2 21 26]