我有这样的专栏-
COL A
product a [Class A], product a [Class C]
product a [Class H], product b [Class A], product b [Class B]
product c [Class C], product d [Class A]想用熊猫来做这个-
Class A Class B Class C Class H
product a product a NaN NaN
product b product b NaN product a
product d NaN product c NaN发布于 2022-08-24 12:23:25
输入:
temp = pd.DataFrame({'COL A':['product a [Class A], product a [Class C]',
'product a [Class H], product b [Class A], product b [Class B]',
'product c [Class C], product d [Class A]']})使用字符串拆分操作,首先拆分“,”获取每一行上的对列表,然后拆分“[”以获取列名和值,并将它们存储在字典中
d = {}
cnt = 0
for s in temp['COL A']:
d[cnt] = {i.split('[')[1][:-1]:i.split('[')[0].strip() for i in s.split(',')}
cnt += 1字典d:
{0: {'Class A': 'product a', 'Class C': 'product a'},
1: {'Class H': 'product a', 'Class A': 'product b', 'Class B': 'product b'},
2: {'Class C': 'product c', 'Class A': 'product d'}}然后从字典中创建一个新的数据格式:(使用transpose将A类和B类作为列,然后使用sort_index对列名排序)
df = pd.DataFrame(d).transpose().sort_index(axis=1)输出:
Class A Class B Class C Class H
0 product a NaN product a NaN
1 product b product b NaN product a
2 product d NaN product c NaN发布于 2022-08-24 12:45:20
下面是使用str.split和explode将数据带入更规范形式的另一种方法,然后使用正则表达式(根据需要修改它)来extract产品名称和类:
([\w\s]+):字母序列与whitespace\[(.+)\]:在[]之间的关系
最后一步是获得所需格式的基本pivot。
df2 = (df["COL A"].str.split(",") # split into lists
.explode() # explode lists into rows
.str.extract("([\w\s]+)\[(.+)\]") # extract info
.pivot(values=0, columns=1)) # pivot结果:
1 Class A Class B Class C Class H
0 product a NaN product a NaN
1 product b product b NaN product a
2 product d NaN product c NaNhttps://stackoverflow.com/questions/73452492
复制相似问题