Python 当数据共享一列值时，是否有方法将数据压缩到数据帧的一行中？

我有一个有几千行的数据框。 DF保存组织内各单位的单位标识符和响应时间。 它是在DF中构造的，具有列[“事件#”，“UnitID”，“第一个UnitInRoute”，“第一个UnitArrived”，“第一个UnitThospital”]

同一事件#有许多不同的行，最后我只希望每个事件#有一行，[“First UnitEnroute”、“First UnitArrived”、“First UnitAtHospital]”由具有相同事件的其他行填充

造成这种情况的原因是某个季度末的账单失败，我们需要知道这些不同的事件是否在不同的单位中传播了这3次。不过，我不需要列出单位，只需要从同一事件的其他行中提取第一个非0值

以下是一些示例数据：

Event#      Unit    First UnitEnroute           First UnitArrived           First UnitAtHospital
2020000394    37    ['1/1/2020', '10:45:34 PM'] ['1/1/2020', '10:48:33 PM'] ['1/1/2020', '11:45:01 PM']
2020000394    38    ['1/1/2020', '10:45:34 PM'] ['1/1/2020', '10:48:33 PM'] ['1/1/2020', '11:45:01 PM']     
2020000394    36    ['1/1/2020', '10:45:34 PM'] ['1/1/2020', '10:48:33 PM'] ['1/1/2020', '11:45:01 PM']     
2020000394    39    ['1/1/2020', '10:45:34 PM'] ['1/1/2020', '10:48:33 PM'] ['1/1/2020', '11:45:01 PM']     
2020000617    58    ['1/2/2020', '12:06:13 PM'] ['1/2/2020', '12:07:39 PM'] ['1/2/2020', '12:43:10 PM']     
2020000849    74    ['1/2/2020', '6:42:19 PM']  ['1/2/2020', '6:53:53 PM']  ['1/2/2020', '7:28:32 PM']      
2020000849    75    ['0']                       ['0']                       ['0']
2020000927    81    ['0']                       ['0']                       ['0']
2020000927    80    ['0']                       ['0']                       ['0']
2020000997    86    ['0']                       ['0']                       ['0']
2020000997    87    ['0']                       ['0']                       ['0']
2020001218    99    ['1/3/2020', '11:50:39 AM'] ['1/3/2020', '11:52:40 AM'] ['1/3/2020', '12:29:37 PM']     
2020001218    98    ['0']                       ['1/3/2020', '11:52:40 AM'] ['0']
2020001255    102   ['1/3/2020', '12:44:30 PM'] ['0']                       ['0']
2020001255    103   ['1/3/2020', '12:40:19 PM'] ['0']                       ['0']
2020001258    98    ['1/3/2020', '12:49:00 PM'] ['1/3/2020', '12:57:22 PM'] ['1/3/2020', '1:39:03 PM']      
2020001258    103   ['0']                       ['0']                       ['0']
2020001258    104   ['0']                       ['0']                       ['0']
2020001258    105   ['0']                       ['0']                       ['0']

这就是我尝试过的：

强制遍历每一行以查找非0值，然后将该值附加到该行

 for row in DF:
     compare = list()
     for i in DF:
         if i[0] == row[0]:
             addition = list(i)
             compare = compare.append(addition)
             print("Compare: {}".format(compare))
             return compare

     for el in row.index:
         whatisit = row[el]
         if whatisit == 0:
             for item in compare.index:
                 if item[el] == 0:
                     return
                 else:
                     replacement = item[el]
                     print("Replacement: {}".format(replacement))
                     return replacement
         row[el] = replacement
         return DF

使用pandas.groupby（）.fillna（）对具有相似事件的行进行分组，并使用时间进行回填。 --我没有得到任何可以接受的结果。大多数时候，它会把我的DF切碎成一些不稳定的形状。我可能误解了如何使用这个

---

任何方向都是感激的，很抱歉，如果这已经被张贴之前，我花了大量的时间寻找一个潜在的答案。我想我还没有完全发展出直觉，我需要看到任何代码，看看如何将其应用到我的项目中。我不是一个专业的开发人员，我更多的是一个动手做繁重事情的员工哈哈。

这里有一个，IIUC

from io import StringIO
import pandas as pd

# create data frame
df = pd.read_csv(StringIO(data), sep='\s\s+', engine='python')

# drop the column `Unit`
df = df.drop(columns='Unit')

# re-shape
df = df.melt(id_vars='Event#', var_name='first_unit', value_name='timestamp')

# drop timestamp == ['0']
mask = df['timestamp'].astype(str) != "['0']"
df = df[mask]

# drop duplicates
df = df.drop_duplicates()

# get min value for each group -- and re-shape
df = (df.groupby(['Event#', 'first_unit'])['timestamp'].min()
      .unstack(level='first_unit')
      .reset_index()
     )

print(df)

first_unit      Event#            First UnitArrived  \
0           2020000394  ['1/1/2020', '10:48:33 PM']   
1           2020000617  ['1/2/2020', '12:07:39 PM']   
2           2020000849   ['1/2/2020', '6:53:53 PM']   
3           2020001218  ['1/3/2020', '11:52:40 AM']   
4           2020001255                          NaN   
5           2020001258  ['1/3/2020', '12:57:22 PM']   

first_unit         First UnitAtHospital            First UnitEnroute  
0           ['1/1/2020', '11:45:01 PM']  ['1/1/2020', '10:45:34 PM']  
1           ['1/2/2020', '12:43:10 PM']  ['1/2/2020', '12:06:13 PM']  
2            ['1/2/2020', '7:28:32 PM']   ['1/2/2020', '6:42:19 PM']  
3           ['1/3/2020', '12:29:37 PM']  ['1/3/2020', '11:50:39 AM']  
4                                   NaN  ['1/3/2020', '12:40:19 PM']  
5            ['1/3/2020', '1:39:03 PM']  ['1/3/2020', '12:49:00 PM'] 

以下是原始数据（即用于创建数据框）：

---

这是一个有，IIUC

from io import StringIO
import pandas as pd

# create data frame
df = pd.read_csv(StringIO(data), sep='\s\s+', engine='python')

# drop the column `Unit`
df = df.drop(columns='Unit')

# re-shape
df = df.melt(id_vars='Event#', var_name='first_unit', value_name='timestamp')

# drop timestamp == ['0']
mask = df['timestamp'].astype(str) != "['0']"
df = df[mask]

# drop duplicates
df = df.drop_duplicates()

# get min value for each group -- and re-shape
df = (df.groupby(['Event#', 'first_unit'])['timestamp'].min()
      .unstack(level='first_unit')
      .reset_index()
     )

print(df)

first_unit      Event#            First UnitArrived  \
0           2020000394  ['1/1/2020', '10:48:33 PM']   
1           2020000617  ['1/2/2020', '12:07:39 PM']   
2           2020000849   ['1/2/2020', '6:53:53 PM']   
3           2020001218  ['1/3/2020', '11:52:40 AM']   
4           2020001255                          NaN   
5           2020001258  ['1/3/2020', '12:57:22 PM']   

first_unit         First UnitAtHospital            First UnitEnroute  
0           ['1/1/2020', '11:45:01 PM']  ['1/1/2020', '10:45:34 PM']  
1           ['1/2/2020', '12:43:10 PM']  ['1/2/2020', '12:06:13 PM']  
2            ['1/2/2020', '7:28:32 PM']   ['1/2/2020', '6:42:19 PM']  
3           ['1/3/2020', '12:29:37 PM']  ['1/3/2020', '11:50:39 AM']  
4                                   NaN  ['1/3/2020', '12:40:19 PM']  
5            ['1/3/2020', '1:39:03 PM']  ['1/3/2020', '12:49:00 PM'] 

以下是原始数据（即用于创建数据框）：

---

因为您有几千行，我建议分别处理每一列，然后再将它们合并在一起：

df1 = df[ df['First UnitEnroute']!="['0']" ][['Event#', 'First UnitEnroute']]
df1 = df1[~df1.duplicated(['Event#'])]
df2 = df[ df['First UnitArrived']!="['0']" ][['Event#', 'First UnitArrived']]
df2 = df2[~df2.duplicated(['Event#'])]
df3 = df[ df['First UnitAtHospital']!="['0']" ][['Event#', 'First UnitAtHospital']]
df3 = df3[~df3.duplicated(['Event#'])]
df_result = df1.merge(df2, on = 'Event#', how='left').merge(df3, on = 'Event#', how='left')

---

通过这种方式（如果我正确理解了这个问题），您可以找到一个或多个第一个单元统计数据没有时间戳的事件。在您的示例中是event 2020001255，因为您有几千行，所以我建议分别处理每一列并再次将它们合并在一起：

df1 = df[ df['First UnitEnroute']!="['0']" ][['Event#', 'First UnitEnroute']]
df1 = df1[~df1.duplicated(['Event#'])]
df2 = df[ df['First UnitArrived']!="['0']" ][['Event#', 'First UnitArrived']]
df2 = df2[~df2.duplicated(['Event#'])]
df3 = df[ df['First UnitAtHospital']!="['0']" ][['Event#', 'First UnitAtHospital']]
df3 = df3[~df3.duplicated(['Event#'])]
df_result = df1.merge(df2, on = 'Event#', how='left').merge(df3, on = 'Event#', how='left')

---

通过这种方式（如果我正确理解了这个问题），您可以找到一个或多个第一个单元统计数据没有时间戳的事件。您的示例中是事件2020001255

谢谢，这看起来是我需要的。我在

“Event”

上的

df=df.melt（id_vars='Event'，var_name='first_unit'，value_name='timestamp'）

行遇到一个键错误，知道为什么吗？你能在df.columns中执行

'Event.

吗？列名中有嵌入空格吗？KeyError可能意味着您指定为id_var的内容不在df.columns中谢谢，这看起来像是我需要的。我在

“Event”

上的

df=df.melt（id_vars='Event'，var_name='first_unit'，value_name='timestamp'）

行遇到一个键错误，知道为什么吗？你能在df.columns中执行

'Event.

吗？列名中有嵌入空格吗？KeyError可能意味着您指定为id_var的内容不在df列中