Python 将计数器转换为 DataFrame 列

Question

我无法在这里找到特定于我的问题的答案，我想知道我是否可以获得一些帮助（抱歉链接，我还不允许嵌入图像）。

我已将 Counter 对象存储在我的 DataFrame 中，并且还希望将它们作为每个计数元素的列添加到 DataFrame 中。

起始数据

data = {
    "words": ["ABC", "BCDB", "CDE", "F"],
    "stuff": ["abc", "bcda", "cde", "f"]
}
df = pd.DataFrame(data)

Preliminary Data Frame

patternData = {
    "name": ["A", "B", "C", "D", "E", "F"],
    "rex": ["A{1}", "B{1}", "C{1}", "D{1}", "E{1}", "F{1}"]
}
patterns = pd.DataFrame(patternData)

Pattern DataFrame

def countFound(ps):
    result = Counter()
    for index, row in patterns.iterrows():
        findName = row['name']
        findRex = row['rex']
        found = re.findall(findRex, ps)
        if (len(found) > 0):
            result.update({findName:len(found)})
    return result

df['found'] = df['words'].apply(lambda x: countFound(x))

Found DataFrame

Desired Results

words	stuff	found	A	B	C	D	E	F
ABC	acb	`{'A': 1, 'B': 1, 'C': 1}`	1	1	1	0	0	0
BCD	bcd	`{'B': 1, 'C': 1, 'D': 1}`	0	2	1	1	0	0
CDE	cde	`{'C': 1, 'D': 1, 'E': 1}`	0	0	1	1	1	0
F	f	`{'F': 1}`	0	0	0	0	0	1

Answer 1

A Counter 的行为很像字典。在字典列表上调用 pd.DataFrame 将为您提供计数值矩阵：

found = df['words'].apply(countFound).to_list()
pd.concat([
    df.assign(found=found),
    pd.DataFrame(found).fillna(0).astype("int")
], axis=1)

Answer 2

您可以使用 json_normalize:

df.join(pd.json_normalize(df['found']).fillna(0, downcast='infer'))

输出：

  words stuff                     found  A  B  C  D  E  F
0   ABC   abc  {'A': 1, 'B': 1, 'C': 1}  1  1  1  0  0  0
1  BCDB  bcda  {'B': 2, 'C': 1, 'D': 1}  0  2  1  1  0  0
2   CDE   cde  {'C': 1, 'D': 1, 'E': 1}  0  0  1  1  1  0
3     F     f                  {'F': 1}  0  0  0  0  0  1

您也可以不使用自定义函数直接获取列。为此，请使用带有命名捕获组和 str.extractall:

的动态制作的正则表达式

regex = ('(?P<'+patterns['name']+'>'+patterns['rex']+')').str.cat(sep='|')
# (?P<A>A{1})|(?P<B>B{1})|(?P<C>C{1})|(?P<D>D{1})|(?P<E>E{1})|(?P<F>F{1})

df2 = df.join(df
 ['words']
 .str.extractall(regex)
 .groupby(level=0).count()
 )

或者没有命名捕获组的变体，稍后设置列名：

regex = ('('+patterns['rex']+')').str.cat(sep='|')
# (A{1})|(B{1})|(C{1})|(D{1})|(E{1})|(F{1})

print(df.join(df
 ['words']
 .str.extractall(regex)
 .set_axis(patterns['name'], axis=1)
 .groupby(level=0).count()
 ))

输出：

  words stuff  A  B  C  D  E  F
0   ABC   abc  1  1  1  0  0  0
1  BCDB  bcda  0  2  1  1  0  0
2   CDE   cde  0  0  1  1  1  0
3     F     f  0  0  0  0  0  1

Python 将计数器转换为 DataFrame 列

Python Convert Counters into DataFrame Columns

python

counter

dataframe

pandas