python - 如何更改 NaN 值之前和之后的值？

Question

如何更改 NaN 值之前和之后的值？
示例数据帧:

df = pd.DataFrame({"a":[1, np.nan, 3, 5 , 6,3,np.nan],"b":[np.nan, 2, np.nan, 
7,2,1,1],"c":[np.nan,4, 8, 9,7,6,5]})

预期输出:

df_expected = pd.DataFrame({"a":[0, np.nan, 0, 5 , 6,0,np.nan],"b":[np.nan, 0, np.nan, 
0,2,1,1],"c":[np.nan,0, 8, 9,7,6,5]})

提前致谢!

Answer 1

使用 DataFrame.mask 由 | 链接按位 OR值上下移动 DataFrame.shift 仅过滤 non NaN s by 链接 &为 bitwise AND :

df = df.mask((df.shift().isna() | df.shift(-1).isna()) & df.notna(), 0)
print (df)
     a    b    c
0  0.0  NaN  NaN
1  NaN  0.0  0.0
2  0.0  NaN  8.0
3  5.0  0.0  9.0
4  6.0  2.0  7.0
5  0.0  1.0  6.0
6  NaN  0.0  0.0

另一个解决方案 fill_value具有任何值并由 | 链接应该用于不使用 notna面具:

df = df.mask(df.shift(fill_value=1).isna() | df.shift(-1, fill_value=1).isna(), 0)
print (df)
     a    b    c
0  0.0  NaN  NaN
1  NaN  0.0  0.0
2  0.0  NaN  8.0
3  5.0  0.0  9.0
4  6.0  2.0  7.0
5  0.0  1.0  6.0
6  NaN  1.0  5.0