CODEWITHSUNDEEP
pythonpandas
JD Long
JD Long

Reputation: 60746

Take a Pandas Series where each element is a DataFrame and combine them to one big DataFrame

I have a Pandas Series where each element of the series is a one row Pandas DataFrame which I would like to append together into one big DataFrame. For example:

import pandas as pd
mySeries = pd.Series( numpy.arange(start=1, stop=5, step=1) )

def myFun(val):
    return pd.DataFrame( { 'square' : [val**2],
                           'cube' :  [val**3] } )
## returns a Pandas Series where each element is a single row dataframe
myResult = mySeries.apply(myFun)

so how do I take myResult and combine all the little dataframes into one big dataframe?

Upvotes: 1

Views: 747

Answers (3)

waitingkuo
waitingkuo

Reputation: 93804

concat them:

In [58]: pd.concat(myResult).reset_index(drop=True)
Out[58]: 
   cube  square
0     1       1
1     8       4
2    27       9
3    64      16

Since the original indexes are all 0, I also reset them.

Upvotes: 2

unutbu
unutbu

Reputation: 879611

import pandas as pd
import numpy as np
mySeries = pd.Series(np.arange(start=1, stop=5, step=1))


def myFun(val):
    return pd.Series([val ** 2, val ** 3], index=['square', 'cube'])

myResult = mySeries.apply(myFun)
print(myResult)

yields

   square  cube
0       1     1
1       4     8
2       9    27
3      16    64

Upvotes: 3

Rutger Kassies
Rutger Kassies

Reputation: 64443

Its seems overly complicated, although you probably posted a simplified example. Creating a new Series for each row creates a lot of overhead. This for example is over 200 times faster (for n=500) on my machine:

meResult = pd.DataFrame({'square': mySeries**2,'cube': mySeries**3})

Upvotes: 1

Related Questions