Question

我有以下示例数据框：

df_1：

from datetime import datetime
import pandas as pd

>>> df_1 = pd.DataFrame( 
 {"SVDiscrep_Merge": ["2081916SAN", "2081242DFW", "2081248ORD","20874CLE", "2081740DEN"],
 "RON_DATE": [datetime(2017,6,1), datetime(2017,6,4), datetime(2017,6,6), datetime(2017,6,7), datetime(2017,6,8)],
 "Next SV1 Date": [datetime(2017,6,4), datetime(2017,6,6), datetime(2017,6,7), datetime(2017,6,8), datetime(2017, 6, 18)]})

>>> df_1

SVDiscrep_Merge     RON_DATE    Next SV1 Date
2081916SAN          6/1/2017    6/4/2017    
2081242DFW          6/4/2017    6/6/2017
2081248ORD          6/6/2017    6/7/2017
20874CLE            6/7/2017    6/8/2017
2081740DEN          6/8/2017    6/18/2017

df_2：

>>> df_2 = pd.DataFrame( 
 {"SVDiscrep_Merge": ["2081916SAN", "2081916SAN", "2081916SAN","2081740DEN"],
 "REPORT_DT": [datetime(2017,6,1), datetime(2017,6,3), datetime(2017,6,4), datetime(2017,6,9)],
 "ColA": ["A", "B", "C", "D"]})

>>> df_2

SVDiscrep_Merge REPORT_DT   ColA
2081916SAN      6/1/2017    A
2081916SAN      6/3/2017    B
2081916SAN      6/4/2017    C
2081740DEN      6/9/2017    D

我想采用以下逻辑：

如果（且仅当）df_2在两个数据框架中相等且

df_1

SVDiscrep_Merge

REPORT_DT列是＆gt; = RON_DATE中的日期和＆lt; Next SV1 Date中df_1的日期。

这是我想要的输出：

SVDiscrep_Merge     RON_DATE    Next SV1 Date  ColA 
2081916SAN          6/1/2017    6/4/2017       A    
2081916SAN          6/4/2017    6/6/2017       B
2081916SAN          6/6/2017    6/7/2017
2081242DFW          6/4/2017    6/6/2017    
2081248ORD          6/6/2017    6/7/2017
20874CLE            6/7/2017    6/8/2017
2081740DEN          6/8/2017    6/18/2017      D

我知道如果我没有那个日期逻辑，如何在python代码中进行合并...但是使用那个日期逻辑（在搜索Google之后）我不知所措。

Answer 1

您可以在SVDiscrep_Merge上保留合并，然后使用以下布尔掩码过滤结果：

mask = (((result['RON_DATE'] <= result['REPORT_DT']) 
         & (result['REPORT_DT'] < result['Next SV1 Date'])) 
        | pd.isnull(result['REPORT_DT']))

import datetime as DT 
import pandas as pd

df_1 = pd.DataFrame( 
 {"SVDiscrep_Merge": ["2081916SAN", "2081242DFW", "2081248ORD","20874CLE", "2081740DEN"],
 "RON_DATE": [DT.datetime(2017,6,1), DT.datetime(2017,6,4), DT.datetime(2017,6,6), DT.datetime(2017,6,7), DT.datetime(2017,6,8)],
 "Next SV1 Date": [DT.datetime(2017,6,4), DT.datetime(2017,6,6), DT.datetime(2017,6,7), DT.datetime(2017,6,8), DT.datetime(2017, 6, 18)]})

df_2 = pd.DataFrame( 
 {"SVDiscrep_Merge": ["2081916SAN", "2081916SAN", "2081916SAN","2081740DEN"],
 "REPORT_DT": [DT.datetime(2017,6,1), DT.datetime(2017,6,3), DT.datetime(2017,6,4), DT.datetime(2017,6,9)],
 "ColA": ["A", "B", "C", "D"]})

result = pd.merge(df_1, df_2, on='SVDiscrep_Merge',  how='left')
mask = (((result['RON_DATE'] <= result['REPORT_DT']) 
         & (result['REPORT_DT'] < result['Next SV1 Date'])) 
        | pd.isnull(result['REPORT_DT']))
result = result.loc[mask].drop('REPORT_DT', axis=1)
print(result)

产量

  Next SV1 Date   RON_DATE SVDiscrep_Merge ColA
0    2017-06-04 2017-06-01      2081916SAN    A
1    2017-06-04 2017-06-01      2081916SAN    B
3    2017-06-06 2017-06-04      2081242DFW  NaN
4    2017-06-07 2017-06-06      2081248ORD  NaN
5    2017-06-08 2017-06-07        20874CLE  NaN
6    2017-06-18 2017-06-08      2081740DEN    D

这不是您发布的理想结果，但它与逻辑描述一致。

Pandas Dataframe合并在包括条件合并的2列上：如果df_2中的日期在df_1中的两个其他日期之间

1 个答案: