在我的代码中，我使用了sklearn的KMeans算法。当我执行代码时，出现了这样的错误：“KMeans对象没有属性’labels_’”

Traceback (most recent call last): File ".\kmeans.py", line 56, in <module>   np.unique(km.labels_, return_counts=True)AttributeError: 'KMeans' object has no attribute 'labels_'

这是我的代码：

import pandas as pdsimport nltk,re,stringfrom nltk.probability import FreqDistfrom collections import defaultdictfrom nltk.tokenize import sent_tokenize, word_tokenize, RegexpTokenizerfrom nltk.tokenize.punkt import PunktSentenceTokenizerfrom nltk.corpus import stopwordsfrom string import punctuationfrom heapq import nlargest# 导入并实例化CountVectorizerfrom sklearn.feature_extraction.text import CountVectorizervect = CountVectorizer()    from sklearn.feature_extraction.text import TfidfVectorizervectorizer = TfidfVectorizer(ngram_range=(1,2),max_df=0.5, min_df=2,stop_words='english')train_X = vectorizer.fit_transform(x)  from sklearn.cluster import KMeansimport sklearn.cluster.k_means_km = KMeans(n_clusters=3, init='k-means++', max_iter=100, n_init=1,   verbose=True)import numpy as npnp.unique(km.labels_, return_counts=True)text = {}for i,cluster in enumerate(km.labels_):    oneDocument = X[i]         if cluster not in text.keys():        text[cluster] = oneDocument    else:        text[cluster] += oneDocument        _stopwords = set(stopwords.words('english')+ list(punctuation))keywords = {}counts = {}for cluster in range(3):    word_sent =  word_tokenize(text[cluster].lower())    word_sent = [word for word in word_sent if word not in _stopwords]    freq = FreqDist(word_sent)    keywords[cluster] =  nlargest(100, freq, key=freq.get)    counts[cluster] = frequnique_keys={}for cluster in range(3):    other_clusters = list(set(range(3))-set([cluster]))    keys_other_clusters =     set(keywords[other_clusters[0]]).union(set(keywords[other_clusters[1]]))    unique=set(keywords[cluster])-keys_other_clusters    unique_keys[cluster]= nlargest(100, unique, key=counts[cluster].get)#print(unique_keys)print(keywords)

为了获取关键词聚类。我尝试解决这个问题..但我不知道问题出在哪里..

回答：

你需要先对KMeans对象进行拟合操作，它才会有标签属性：

如果没有拟合，它会抛出错误：

from sklearn.cluster import KMeanskm = KMeans()print(km.labels_)>>>AttributeError: 'KMeans' object has no attribute 'labels_'

拟合之后：

from sklearn.cluster import KMeansimport numpy as npkm = KMeans()X = np.random.rand(100, 2)km.fit(X)print(km.labels_)>>>[1 6 7 4 6 6 7 5 6 0 0 7 3 4 5 7 5 0 3 4 0 6 1 6 7 5 4 3 4 2 1 2 1 4 6 3 6 1 7 6 6 7 4 1 1 0 4 2 5 0 6 3 1 0 7 6 2 7 7 5 2 7 7 3 2 1 2 2 4 7 5 3 2 65 1 6 2 4 2 3 2 2 2 1 2 0 5 7 2 4 4 5 4 4 1 1 4 5 0]