我要搜索等于id1或id2的列ID,然后在F1中添加值为col3的列。还有NAN。
d = {'id1': ["ABC","ANB","ATB","BTP"],'id2':["XXX","YYY","ZZZ","TTT"], 'Name': ["A1","A2","A3","A4"]}
F1 = pd.DataFrame(data=d)
d = {'id': ["ABC","ANB","ZZZ"], 'col3': [0,1,1]}
F2 = pd.DataFrame(data=d)
我做了这行代码,但是没有给出预期的结果。
pd.concat([F1.merge(F2, left_on='id1', right_on='id'),F1.merge(F2, left_on='id2', right_on='id')], axis=0).drop(['Name','id'], axis=1)
预期输出如下图所示。enter image description here
答案 0 :(得分:1)
对Series.map
和DataFrame.set_index
创建的Series
的两列都使用双Series.fillna
,以替换缺失的值:
s = F2.set_index('id')['col3']
F1['col3'] = F1['id1'].map(s).fillna(F1['id2'].map(s))
print (F1)
id1 id2 Name col3
0 ABC XXX A1 0.0
1 ANB YYY A2 1.0
2 ATB ZZZ A3 1.0
3 BTP TTT A4 NaN
详细信息:
print (F1['id1'].map(s))
0 0.0
1 1.0
2 NaN
3 NaN
Name: id1, dtype: float64
print (F1['id2'].map(s))
0 NaN
1 NaN
2 1.0
3 NaN
Name: id2, dtype: float64
print(F1['id1'].map(s).fillna(F1['id2'].map(s)))
0 0.0
1 1.0
2 1.0
3 NaN
Name: id1, dtype: float64
您的解决方案应使用左联接和fillna
进行更改:
a = F1.merge(F2, left_on='id1', right_on='id', how='left')['col3']
b = F1.merge(F2, left_on='id2', right_on='id', how='left')['col3']
F1['col3'] = a.fillna(b)
print (F1)
id1 id2 Name col3
0 ABC XXX A1 0.0
1 ANB YYY A2 1.0
2 ATB ZZZ A3 1.0
3 BTP TTT A4 NaN
如果最后一列中需要整数,则最后一个需要使用Int64
进行转换的熊猫0.24 +:
F1['col3'] = F1['id1'].astype('Int64')
print (F1)
id1 id2 Name col3
0 ABC XXX A1 0
1 ANB YYY A2 1
2 ATB ZZZ A3 1
3 BTP TTT A4 NaN