聚类后原始数据集的重复

Question

from sklearn.feature_extraction.text import TfidfVectorizer

tfidf_vectorizer = TfidfVectorizer(max_df=0.08, max_features=200,
                                 min_df=0.02, stop_words='english',
                                 use_idf=True, ngram_range=(1,3),tokenizer = tokenize_only_subject, analyzer='word')

tfidf_vectorizer.fit(enron_data["headers.Subject"])


tfidf_matrix_subject = tfidf_vectorizer.fit_transform(enron_data["headers.Subject"])


print "\n\nshape of tfidf :\t",(tfidf_matrix_subject.shape)

terms_subject = tfidf_vectorizer.get_feature_names()
print "\n Feature's selected by machine from tdifd for Subject :\t",terms_subject

x =tfidf_matrix_subject.toarray()
#
#######################################################################################
from sklearn.metrics.pairwise import cosine_similarity
distance = 1 - cosine_similarity(tfidf_matrix_subject)

print "+++distance\t:",distance[:5]


from sklearn.cluster import KMeans

num_clusters = 4

km = KMeans(n_clusters=num_clusters)


print ":",km.fit_transform(tfidf_matrix_subject).shape


centroids = km.cluster_centers_
labels = km.labels_

print "Centroid is:\t",centroids
print "Labels is :\t",labels
n_clusters_ = km.labels_
print "++++++++++++++++++++++++++++++++++++++++++++++\n",n_clusters_
enron_cls = { 'enron_data_body': enron_data["body"],'enron_data_Subject': enron_data["headers.Subject"],'_id_':enron_data["_id"],"Date":enron_data["Date"],'cluster_': n_clusters_}

frame = pd.DataFrame(enron_cls, index = [n_clusters_] , columns = ['_id_','enron_data_body','enron_data_Subject','Date','cluster_'])

print frame.head()
frame.to_csv("errror.csv")

我需要指导或帮助。 它给出了repeat值；例如，原始数据集的第四行的重复次数与使用cluster的数据集计数的重复次数相同。 我想对每一行进行聚类，而不是重复原始数据集。你知道吗