Pandas：基于公共列连接两个数据帧的最佳方法

Question

I know this is a basic question. But, please hear me out.

I have below dataframes:

In [722]: m1
Out[722]: 
   Person_id  Evidence_14 Feature_14
0        100         90.0       True
1        101          NaN        NaN
2        102         91.0       True
3        103          NaN        NaN
4        104         94.0       True
5        105          NaN        NaN
6        106          NaN        NaN

In [721]: m3
Out[721]: 
   Person_id  Evidence_14 Feature_14
0        100          NaN        NaN
1        101         99.0      False
2        102          NaN        NaN
3        103         95.0      False
4        104          NaN        NaN
5        105          NaN        NaN
6        106         93.0      False

Expected Output:

In [734]: z
Out[734]: 
   Person_id  Evidence_14 Feature_14
0        100         90.0       True
1        101         99.0      False
2        102         91.0       True
3        103         95.0      False
4        104         94.0       True
5        105          NaN        NaN
6        106         93.0      False

I am able to solve this like below:

In [725]: z = m1.merge(m3, on='Person_id')
In [728]: z['Evidence_14'] = z.Evidence_14_x.combine_first(z.Evidence_14_y)
In [731]: z['Feature_14'] = z.Feature_14_x.combine_first(z.Feature_14_y)
In [733]: z.drop(['Evidence_14_x', 'Evidence_14_y', 'Feature_14_x', 'Feature_14_y'], 1, inplace=True)

In [734]: z
Out[734]: 
   Person_id  Evidence_14 Feature_14
0        100         90.0       True
1        101         99.0      False
2        102         91.0       True
3        103         95.0      False
4        104         94.0       True
5        105          NaN        NaN
6        106         93.0      False

But, is there a cleaner/better way to do this? Am I missing something very obvious?

Answer 1

If columns names matching and need match by Person_id values use:

m = m1.set_index('Person_id').combine_first(m2.set_index('Person_id')).reset_index()

If index values are same and also Person_id are same in both DataFrames solution should be simplify by matching with original index values:

m = m1.combine_first(m2)

Answer 2

As Person_id uniquely define related rows in m1 and m3, you have to use set_index. Look at this:

import pandas as pd

df1 = pd.DataFrame({'id':[11, 22, 33,44],'A': [None, 0, 17, None], 'B': [None, 4, 19,None]})
df2 = pd.DataFrame({'id':[111, 222], 'A': [9999, 9999], 'B': [7777, 7777]})

# df1 = df1.set_index('id')
# df2 = df2.set_index('id')

df1.combine_first(df2)


Out[32]: 
   id       A       B
0  11  9999.0  7777.0
1  22     0.0     4.0
2  33    17.0    19.0
3  44     NaN     NaN

if you dont use set_index the first value of A will be changed even if it's id is 11 in df1 and 111 in df2 (different id)

Also note that if you use set_index, a non existing id in m1 will be Added to the result.