Question

我有一个包含多个列的数据框，另一个包含其他多个列的较大数据框。

df1中的匹配列是symbol，m_date df2中的匹配列是符号，日期

我想获取从df2到df1的其他列的值，例如

if m_date = date matches for a particular symbol, then copy the other columns 
if m_date != date for a particular symbol, then we find the closest date to m_date in df2 and use their values to be added to the df1

我尝试了

merged_left = pd.merge(left=df31, right=df2, how='left', left_on=['symbol','m_date'], right_on=['symbol','date'])

这将合并文件，但是不可用的日期具有对应的值为空白。任何人都可以建议如何实现相同的

谢谢

Answer 1

使用merge_asof，您可以使用by先在符号上合并，然后再在列日期on上查找nearest。日期需要为日期时间并进行排序。

# some data similar to yours but simplified
df1 = pd.DataFrame({'symbol': {0: 'A', 1: 'A', 2: 'A', 3: 'A'},
                     'var1': {0: 34, 1: 45, 2: 43, 3: 67},
                     'm_date': {0: '11/25/19', 1: '8/14/19', 2: '5/14/19', 3: '2/20/19'}})
df2 = pd.DataFrame({'symbol': {0: 'A', 1: 'A', 2: 'A', 3: 'A', 4: 'A'},
                    'date': {0: '1/2/19', 1: '5/3/19', 2: '8/4/19', 3: '1/5/20', 4: '1/8/20'},
                    'Per1d': {0: 1, 1: 3, 2: 5, 3: 8, 4: 6},})

# create a column with the dates as datetime
df1['date_'] = pd.to_datetime(df1['m_date'])
df2['date_'] = pd.to_datetime(df2['date'])

# merge_asof
df3 = (pd.merge_asof(df1.sort_values('date_'), 
                     df2.sort_values('date_'), 
                     by=['symbol'], on=['date_'], 
                     direction='nearest')
      )
print (df3)
  symbol  var1    m_date      date_    date  Per1d
0      A    67   2/20/19 2019-02-20  1/2/19      1
1      A    43   5/14/19 2019-05-14  5/3/19      3
2      A    45   8/14/19 2019-08-14  8/4/19      5
3      A    34  11/25/19 2019-11-25  1/5/20      8

根据查找范围内最接近的日期合并列

1 个答案: