Question

我的数据框如下所示。

请注意，我们只有F-P-F模式。即我们的数据只会有F-P-F模式。

    ID  Status  Date    Duration
0   1   F   2018-06-22  nan
1   1   P   2018-08-22  61.00
2   1   F   2018-10-22  61.00
3   3   F   2018-11-20  nan
4   3   P   2018-12-20  30.00
5   3   F   2019-03-20  90.00
6   4   F   2018-06-10  nan
7   4   P   2018-08-10  61.00
8   4   F   2018-12-10  122.00
9   7   F   2018-04-10  nan
10  7   P   2018-08-10  122.00
11  7   F   2018-11-10  92.00
12  7   P   2019-08-10  273.00
13  7   F   2019-10-10  61.00

我想从上面的数据框中准备下面的数据框。

ID        F_P_Duration    F_F_Duration                          
1         61.0            122.0
3         30.0            120.0
4         61.0            183.0
7_1       122.0           214.0
7_2       273.0           334.0

其中F_P_Duration是从F到P的天数

F_F_Duration是该ID的F-P-F模式中从F到F的天数

Answer 1

根据ID列，您似乎只是从一行中获取持续时间，还是与前一行求和。真正的诡计在于安排和标记。我认为下面的代码应该很容易解释。

# imports
import numpy as np
import pandas as pd

# Setup the data and the DataFrame.
data = [[1, 'F', '2018-06-22', np.nan],
        [1, 'P', '2018-08-22', 61.00],
        [1, 'F', '2018-10-22', 61.00],
        [3, 'F', '2018-11-20', np.nan],
        [3, 'P', '2018-12-20', 30.00],
        [3, 'F', '2019-03-20', 90.00],
        [4, 'F', '2018-06-10', np.nan],
        [4, 'P', '2018-08-10', 61.00],
        [4, 'F', '2018-12-10', 122.00],
        [7, 'F', '2018-04-10', np.nan],
        [7, 'P', '2018-08-10', 122.00],
        [7, 'F', '2018-11-10', 92.00],
        [7, 'P', '2019-08-10', 273.00],
        [7, 'F', '2019-10-10', 61.00]]
df = pd.DataFrame(data=data, columns=['ID', 'Status', 'Date', 'Duration'])

# Add a helper column for summing F_F durations.
df['DurShiftSum'] = df['Duration'] + df['Duration'].shift(1)
# F_P duration just appears to be the duration at P.
df.loc[df['Status']=='P', 'F_P_Duration'] = df.loc[df['Status']=='P', 'Duration']
# F_F durations is the F duration plus the previous P duration.
df.loc[(df['Status']=='F')&(df['Duration'].notnull()), 'F_F_Duration'] = 
df.loc[(df['Status']=='F')&(df['Duration'].notnull()), 'DurShiftSum']
# Compress the DataFrame and drop unneeded columns.
df['F_F_Duration'] = df['F_F_Duration'].fillna(method='bfill', limit=1)
df = df.dropna(subset=['F_P_Duration'])
df = df.drop(labels=['Date', 'Duration', 'DurShiftSum'], axis=1)

# An unfortunate for-loop through the unique IDs.
# If your dataset is very big this might not be ideal.
df['ID'] = df['ID'].astype(str)
for xid in df['ID'].unique():
    if len(df.loc[df['ID']==xid]) > 1:
        len_frame = len(df.loc[df['ID']==xid])
        new_ids = [xid+f'_{i}' for i in range(1, len_frame+1)]
        df.loc[df['ID']==xid, 'ID'] = new_ids

根据您提供的示例数据，我将与您期望的结果相匹配。我要做的就是创建一个辅助列，将两个相邻的行相加，将适当的值传输到F_P和F_F列，然后进行清理和格式化。

根据其他列的序列上的某些条件进行分组和

1 个答案: