我在 this dataset of Jeopardy questions 上运行谱聚类,而且我在处理数据时遇到了这个令人沮丧的问题。请注意,我只是将“问题”列中的所有值聚类。
当我在数据集上运行双聚类时,显然出现了“被零除”的 ValueError。
/usr/local/lib/python3.6/dist-packages/sklearn/cluster/bicluster.py:38: RuntimeWarning: divide by zero encountered in true_divide
row_diag = np.asarray(1.0 / np.sqrt(X.sum(axis=1))).squeeze()
/usr/local/lib/python3.6/dist-packages/sklearn/cluster/bicluster.py:286: RuntimeWarning: invalid value encountered in multiply
z = np.vstack((row_diag[:, np.newaxis] * u,
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
...
ValueError: Input contains NaN, infinity or a value too large for dtype('float64').
该错误显然表明我的数据中潜伏着一个 NaN 或无限值(这只是问题的单数列)。我正在处理的完全是文本数据,我已经尝试过大多数用于过滤 NaN 和 inf 的 NumPy 和 Pandas 函数,以及 Stack Overflow 上的许多解决方案。我找不到。
为了确保我的代码没有错误,同样的事情在二十个新闻组数据集上完美运行。
Here's the code on Kaggle if you want to run it and see for yourself.但是,为了以防万一 SO 的政策禁止这样做,下面是代码简述:
dat = pd.DataFrame(pd.read_csv('../input/jarchive_cleaned.csv'))
qlist = []
def cleanhtml(raw_html):
cleanr = re.compile('<.*?>')
cleantext = re.sub(cleanr, '', raw_html)
return cleantext
for row in dat.iterrows():
txt = row[1]['text'].lower()
txt = cleanhtml(txt)
txt = re.sub(r'[^a-z ]',"",txt)
txt = re.sub(r' ',' ',txt)
# txt = ' '.join([stem(w) for w in txt.split(" ")])
qlist.append([txt,row[1]['answer'],row[1]['category']])
print(qlist[:10])
swords = set(stopwords.words('english'))
tv = TfidfVectorizer(stop_words = swords , strip_accents='ascii')
queslst = [q for (q,a,c) in qlist]
qlen = len(set([c for (q,a,c) in qlist]))
mtx = tv.fit_transform(queslst)
cocluster = SpectralCoclustering(n_clusters=qlen, svd_method='arpack', random_state=0) #
t = time()
cocluster.fit(mtx)
最佳答案
一些字符串序列,例如'down out' 导致 TfidfVectorizer()
返回零值。这会导致错误以除以 零 错误开头,从而导致 mtx
稀疏矩阵<中的
这会导致第二个错误。inf
值
作为此问题的变通方法,在 TfidfVectorizer.fit_transform()
创建后,删除此序列或从 mtx
矩阵中删除零矩阵元素,这有点由于稀疏矩阵运算而变得棘手。
我做了第二个解决方案,因为我没有深入到原来的任务中,如下:
swords = set(stopwords.words('english'))
tv = TfidfVectorizer(stop_words = swords , strip_accents='ascii')
queslst = [q for (q,a,c) in qlist]
qlen = len(set([c for (q,a,c) in qlist]))
mtx = tv.fit_transform(queslst)
indices = []
for i,mx in enumerate(mtx):
if np.sum(mx, axis=1) == 0:
indices.append(i)
mask = np.ones(mtx.shape[0], dtype=bool)
mask[indices] = False
mtx = mtx[mask]
cocluster = SpectralCoclustering(n_clusters=qlen, svd_method='arpack', random_state=0) #
t = time()
cocluster.fit(mtx)
终于成功了。希望对您有所帮助,祝您好运!
关于python - scikit-learn 谱聚类 : unable to find NaN lurking in data,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/53358270/