如果Nan，熊猫会将列值移动到同一列的第一行

Question

How do I go from this df ,

     c0   c1   c2   c3 
0  5.99  NaN  NaN  NaN  
1   NaN   12  NaN  NaN                
2   NaN  NaN   44  NaN               
3   NaN  NaN  NaN   49

to this df ,

     c0   c1   c2   c3 
0  5.99   12   44   49  

That is,

• Move all non-NaN values into first row
• Keep all values in original column

Per column, there's just 1 non-NaN value.

Answer 1

Use bfill with axis=0 and downcast='infer' + dropna :

df = df.bfill(axis=0, downcast='infer').dropna()

print(df)
     c0  c1  c2  c3
0  5.99  12  44  49

Without downcast='infer' you get all float data types:

print(df.bfill(0).dropna())
     c0    c1    c2    c3
0  5.99  12.0  44.0  49.0