Equivalent 'rep' of R in Pandas dataframe

user_10854520 0
Viewed 1993

I have searched for some similar questions like'equivalent R function rep in Python'.

In R,rep can be used to an array or a dataframe and you can set the parameter each to specify whether you want to repeat every element or repeat the whole list/dataframe.

But in Python, you have to distinguish between array and dataframe.

For an array, np.repeat will repeat each element and np.tile repeat the whole array .

x=['a','b']

np.repeat(x,2)#repeat each element twice
Out[85]: array(['a', 'a', 'b', 'b'], dtype='<U1')

np.tile(x,2)#repeat the whole array twice
Out[86]: array(['a', 'b', 'a', 'b'], dtype='<U1')

For a Pandas dataframe. pd.concat can be used to repeat the whole dataframe:

d=pd.DataFrame({'x':['a','b'],'y':['c','d']})
d
Out[94]: 
   x  y
0  a  c
1  b  d


pd.concat([d]*2)
Out[93]: 
   x  y
0  a  c
1  b  d
0  a  c
1  b  d

My question is how to repeat each row in a pandas dataframe rather repeat it as a whole. The result I want is:

x y
a c
a c
b d 
b d

Anyway, I wish there is a function in Python like'rep' which can be used to both list and dataframe , and also can specify repeat as a whole or repeat each element.

rep pandas dataframe r python
2 Answers

In pandas you can using reindex with np.repeat

d.reindex(np.repeat(df.index.values,2))
   x  y
0  a  c
0  a  c
1  b  d
1  b  d

Or re-build your dataframe

pd.DataFrame(np.repeat(d.values,2,axis=0),columns=d.columns)
   x  y
0  a  c
1  a  c
2  b  d
3  b  d

Also concat wih sort_index 

pd.concat([d]*2).sort_index()
   x  y
0  a  c
0  a  c
1  b  d
1  b  d
user_79645270

You can also use np.repeat with np.arange:

In [183]: d.iloc[np.repeat(np.arange(len(d)), 2)]
Out[183]: 
   x  y
0  a  c
0  a  c
1  b  d
1  b  d
user_58208140
Related