在列max中填充NaN

Question

My data:

df = pd.DataFrame({'a':[2, np.nan, 2, np.nan],
                   'b':[np.nan, np.nan, 3, 3]})

a      b
2      np.nan
np.nan np.nan
2       3
np.nan  3

looking to get

a      b
2      3
2      3
2      3
2      3

Tried forwardfill

df = df.ffill()
a      b
2      np.nan
2      np.nan
2      3
2      3

Tried forward fill but because i had nan as the first row in b , i got nan still in b . How can i go about filling with unique values of each row instead? (the columns all have repeating values or np.nan)

Answer 1

Chain a bfill call:

df.ffill().bfill()

     a    b
0  2.0  3.0
1  2.0  3.0
2  2.0  3.0
3  2.0  3.0

---

To allow pandas to downcast to int s if possible, use

df.ffill().bfill(downcast='infer')

   a  b
0  2  3
1  2  3
2  2  3
3  2  3

---

Another option is fillna with max .

df.fillna(df.max(), downcast='infer')

   a  b
0  2  3
1  2  3
2  2  3
3  2  3

max ignores NaNs, so is a good alternative.