熊猫groupby的非缺失值

Question

I have a df like this.

>>> df
    c1  t1      c2
0  NaN  20     xyz
1  NaN  20  x1y1z1
2  NaN  20     NaN
>>> df.groupby(['t1'], as_index=False).agg(lambda x: ';'.join(x.astype(str)))
  t1           c1              c2
0  20  nan;nan;nan  xyz;x1y1z1;nan

But I don't want to join columns values if it's np.nan

desired o/p

   t1         c1              c2
0  20        NaN        xyz;x1y1z1

How this check can be done here?

Thanks

Answer 1

You could try using dropna :

df.groupby('t1', as_index=False)\
  .agg(lambda x: ';'.join(x.dropna().astype(str)))\
  .replace('',np.nan)

Output:

   t1  c1          c2
0  20 NaN  xyz;x1y1z1