有条件地用来自不同数据框的数据填充pandas列

Question

I have a data frame (df1) that has a column, 'units', populated with either blank nan values or strings.

 df1 id s_type s_name s_unit 1 t1 n1 m2 2 t1 n5 m2 3 t2 n2 NaN 4 t6 n3 each 

I have a second dataframe, (df2), that has similar information, though without ids.

 df2 type name unit t3 n4 cm2 t4 n2 m3 t2 n2 kg t6 n0 Nan 

I am struggling to develop an expression to

1. identify rows in df1 where unit is null AND
2. insert unit value from df2 in to unit column of df1 WHERE
3. df1['type'] matches df2['type'] AND df1['name'] matches df2['name']

In the above frames, the expression would populate the 'unit' column of df1 with the value of 'kg', as 'type' and 'name' both match.

Something similar to:

 df1.loc[df1['unit'].isnull(), 'unit'] = df2['unit'].where( (df1['name'] == df2['name']) & (df1['type'] == df2['type'])) 

Though the above line is producing a "ValueError: Can only compare identically-labeled Series objects."

I have looked through the documentation and other SO questions. and am at a loss. Any help would be much appreciated.

Answer 1

You can use merge with left join and then combine_first or fillna :

df = pd.merge(df1, df2, on=['type','name'], how='left')

df1['unit'] = df1['unit'].combine_first(df['unit_y'])
print (df1)
   id type name  unit
0   1   t1   n1    m2
1   2   t1   n5    m2
2   3   t2   n2    kg
3   4   t6   n3  each

---

df1['unit'] = df1['unit'].fillna(df['unit_y'])
print (df1)
   id type name  unit
0   1   t1   n1    m2
1   2   t1   n5    m2
2   3   t2   n2    kg
3   4   t6   n3  each

Answer 2

You can merge first and then fill na in unit with values from df2.

(
     pd.merge(df1,df2,on=['type','name'],how='left',suffixes=['','_y'])
         .assign(unit=lambda x: x.unit.combine_first(x.unit_y))
         .drop('unit_y',1)
)
Out[301]: 
   id type name  unit
0   1   t1   n1    m2
1   2   t1   n5    m2
2   3   t2   n2    kg
3   4   t6   n3  each