Question

由于tf-idf矢量化器只会在遇到新标签时崩溃，我正试图从我的新输入中删除新标签。如何更新数据框的列值？我在做：

def clean_unseen(dfcol, vectorizer):
    cleanedstring = ""
    for entry in dfcol:
        for word in entry.split():
            if word in vectorizer.vocabulary_:
                cleanedstring = cleanedstring + " " + word
                print(cleanedstring)
        entry = cleanedstring
        cleanedstring = ""
    return dfcol

示例：

tfifgbdf_vect= TfidfVectorizer()
s2 = pd.Series(['the cat', 'awesome xyz', 'f_g_h lol asd'])
tfifgbdf_vect.fit_transform(s2)
s3 = pd.Series(['the dog the awesome xyz', 'xyz lol asd', 'f_g_h lol aha'])
clean_unseen(s3, tfifgbdf_vect)

然而，这将使原始列保持不变：

Output: 
0    the dog the awesome xyz
1                xyz lol asd
2              f_g_h lol aha
dtype: object

Answer 1

由于系列中的单个条目不是对象，因此它始终是深层副本而不是引用，您需要明确更改 -

def clean_unseen(dfcol, vectorizer):
    dfc1 = []
    cleanedstring = ""
    for entry in dfcol:
        for word in entry.split():
            if word in vectorizer.vocabulary_:
                cleanedstring = cleanedstring + " " + word
                #print(cleanedstring)
        #entry = cleanedstring
        dfc1.append(cleanedstring)
        cleanedstring = ""

    return pd.Series(dfc1)

tfifgbdf_vect= TfidfVectorizer()
s2 = pd.Series(['the cat', 'awesome xyz', 'f_g_h lol asd'])
tfifgbdf_vect.fit_transform(s2)
s3 = pd.Series(['the dog the awesome xyz', 'xyz lol asd', 'f_g_h lol aha'])
clean_unseen(s3, tfifgbdf_vect)

Pandas - 迭代列并更新值

1 个答案: