Question

我的数据集如下所示：

Name  : joe
Job   : Crazy Consultant
Hired : 4/12/2011 3:38:55 AM
Stats : crazy, bald head
Pay   : $5000 Monthly

Name  : Matt
Job   : Crazy Receptionist
Hired : 4/12/2014 3:38:55 PM
Stats : crazy, Lots of hair

Name  : Adam
Job   : Crazy Drinker
Hired : 4/12/2017 3:38:55 AM
Stats : crazy, unknown
Term  : 4/12/2017 3:38:55 PM

我读入并获取如下数据：

df = pd.read_csv(r"pathtomycsv.csv", encoding="UTF-16", delimiter='\s+:').transpose()

上述输出:(仅作为示例）

Name      Job                Hired                 Stats                Name      Job                Hired                 Stats
Joe       Crazy Consultant   4/12/2011 3:38:55 AM  crazy, bald head     Matt      Crazy Consultant   4/12/2011 3:38:55 AM  crazy, bald head

最终，我想从上面获取我的数据集，并通过将所有标题组合在一起将其转换为如下所示的数据集，如下所示：

Name      Job                Hired                 Stats                Pay            Term
Joe       Crazy Consultant   4/12/2011 3:38:55 AM  crazy, bald head     $5000 Monthly  N/A
Matt      Crazy Receptionist 4/12/2014 3:38:55 PM  crazy, Lots of hair  N/A            N/A
Adam      Crazy Drinker      4/12/2017 3:38:55 AM  crazy, unknown       N/A            4/12/2017 3:38:55 PM

Answer 1

出现问题是因为您在日期中有更多冒号。使用"\s+:\s+"作为分隔符。（是的，它可以是正则表达式。）

以下代码可用于将您的文件转换为所需的表格。我认为＆＃39;姓名＆＃39;始终是集合中的第一行。

df = pd.read_csv("yourfile", delimiter='\s+:\s+',header=None)
df = df.reset_index()
df['index'][df[0]!='Name'] = np.nan
df['index'] = df['index'].fillna(method='ffill').astype(int)
df.set_index(['index',0])[1].unstack().set_index('Name')
#0                    Hired                 Job            Pay  
#Name                                                            
#joe   4/12/2011 3:38:55 AM    Crazy Consultant  $5000 Monthly   
#Matt  4/12/2014 3:38:55 PM  Crazy Receptionist           None   
#Adam  4/12/2017 3:38:55 AM       Crazy Drinker           None

Answer 2

你可以尝试这样：

import pandas as pd

df = pd.read_csv('file_name',sep='\s+:\s+',header=None).pivot(columns=0, values=1)
df.index = [df.index, df.Name.notnull().cumsum() - 1]
df = df.stack().reset_index(name='val')
df = df.pivot(index='Name', columns=0, values='val')
df

输出：

阅读CSV Transpose pandas

2 个答案: