如何在Panda中逐对象插值

Question

原始数据帧

yyyymm

a	b	价格
1	a	200101	3000
1		a	200102	np.nan
1		a	200103	np.nan
1		a	200104	6000
1		b	200101	np.nan
1		b	200102	np.nan
1		b	200103	np.nan
1		b	200104	3000
2		a	200101	3000
2		a	200102	np.nan
2		a	200103	np.nan
2		a	200104	np.nan

这符合预期：

df = pd.DataFrame({'a': [1,1,1,1,1,1,1,1,2,2,2,2],
'b': list('aaaabbbbaaaa'),
'yyyymm': [200101, 200102, 200103, 200104, 200101, 200102, 200103, 200104,
200101, 200102, 200103, 200104],
'price': [3000,np.NaN,np.NaN,6000,np.NaN,np.NaN,np.NaN,3000,3000,np.NaN,np.NaN,np.NaN]
})
df.groupby(['a', 'b'])['price'].apply(
lambda group: group.interpolate(method='linear', limit=2, limit_area='inside')
)

输出：

0     3000.0
1     4000.0
2     5000.0
3     6000.0
4        NaN
5        NaN
6        NaN
7     3000.0
8     3000.0
9        NaN
10       NaN
11       NaN
Name: price, dtype: float64