在两个数据框中按日期标注

Question

df1
name date
A    14-04-05
A    14-05-08
A    14-08-09
A    15-01-05
B    18-07-05
B    18-08-09
B    18-10-02
C    19-01-03
C    19-02-04
C    19-03-30
D    16-04-01
D    16-08-04

df2
name startdate
A    14-07-07 
B    18-09-09
C    19-03-15
D    16-06-28

在一个数据集中记录所有日期，第二个数据集记录开始日期。

我想把df1的记录与df2的开始日期进行比较。记录'0'，如果它比开始日期早，'1'，如果它是开始日期后的一天。

结果是我想要的

df1
name date     Label startdate
A    14-04-05 0     14-07-07
A    14-05-08 0     14-07-07
A    14-08-09 1     14-07-07
A    15-01-05 1     14-07-07
B    18-07-05 0     18-09-09
B    18-08-09 0     18-09-09
B    18-10-02 1     18-09-09
C    19-01-03 0     19-03-15
C    19-02-04 0     19-03-15
C    19-03-30 1     19-03-15
D    16-04-01 0     16-06-28
D    16-08-04 1     16-06-28

我试着用datetime处理它，但没有成功。

简单的数据集示例

df1 = pd.DataFrame(np.array([['A', '2015-12-21'],['A', '2015-12-22'], ['A', '2015-12-25'], ['B', '2018-01-28'],['B', '2018-02-28'],['B', '2018-03-28']]),
                   columns=['name', 'date'])

df2 = pd.DataFrame(np.array([['A', '2015-12-23'], ['B', '2018-03-01']]),
                   columns=['name', 'startdate'])

谢谢你的阅读

Answer 1

使用 DataFrame.merge 添加新列，然后通过 Series.gt 为大 DataFrame.insert 用于按位置新建列，用于转换为数字列。0,1 是用 Series.view:

df1['date'] = pd.to_datetime(df1['date'])
df2['startdate'] = pd.to_datetime(df2['startdate'])

df = df1.merge(df2, on='name', how='left')
df.insert(2, 'Label', df['date'].gt(df['startdate']).view('i1'))
print (df)
   name       date  Label  startdate
0     A 2014-04-05      0 2014-07-07
1     A 2014-05-08      0 2014-07-07
2     A 2014-08-09      1 2014-07-07
3     A 2015-01-05      1 2014-07-07
4     B 2018-07-05      0 2018-09-09
5     B 2018-08-09      0 2018-09-09
6     B 2018-10-02      1 2018-09-09
7     C 2019-01-03      0 2019-03-15
8     C 2019-02-04      0 2019-03-15
9     C 2019-03-30      1 2019-03-15
10    D 2016-04-01      0 2016-06-28
11    D 2016-08-04      1 2016-06-28

或者：

df1['date'] = pd.to_datetime(df1['date'])
df2['startdate'] = pd.to_datetime(df2['startdate'])

df1['startdate'] = df1['name'].map(df2.set_index('name')['startdate'])
df1.insert(2, 'Label', df1['date'].gt(df1['startdate']).view('i1'))
print (df1)
   name       date  Label  startdate
0     A 2014-04-05      0 2014-07-07
1     A 2014-05-08      0 2014-07-07
2     A 2014-08-09      1 2014-07-07
3     A 2015-01-05      1 2014-07-07
4     B 2018-07-05      0 2018-09-09
5     B 2018-08-09      0 2018-09-09
6     B 2018-10-02      1 2018-09-09
7     C 2019-01-03      0 2019-03-15
8     C 2019-02-04      0 2019-03-15
9     C 2019-03-30      1 2019-03-15
10    D 2016-04-01      0 2016-06-28
11    D 2016-08-04      1 2016-06-28

Answer 2

你可以 map 它。

print (df1.assign(new=(df1["date"]>df1["name"].map(df2.set_index("name")["startdate"])).astype(int),
                  start=df1["name"].map(df2.set_index("name")["startdate"])))

   name      date  new     start
0     A  14-04-05    0  14-07-07
1     A  14-05-08    0  14-07-07
2     A  14-08-09    1  14-07-07
3     A  15-01-05    1  14-07-07
4     B  18-07-05    0  18-09-09
5     B  18-08-09    0  18-09-09
6     B  18-10-02    1  18-09-09
7     C  19-01-03    0  19-03-15
8     C  19-02-04    0  19-03-15
9     C  19-03-30    1  19-03-15
10    D  16-04-01    0  16-06-28
11    D  16-08-04    1  16-06-28