CODEWITHSUNDEEP
pythonpandas
interpolack
interpolack

Reputation: 914

Convert subset of columns to list

I have a pivoted Pandas DataFrame with the following columns:

month | day | hour | a | b | c | d | e | f | g ... z
1       1     1      3   9   0   9   0   3   3     9

What is the most efficient way to turn all rows in columns a through z into a list of their values, and use this new list column instead, in-place? The resulting columns would be:

month | day | hour | list
1       1     1      [3,9,0,9,0,3,3,9 ...]

I could iterate the rows and manually combine a through z into many lists, and then delete the unnecessary columns afterward, but there may be a more straightforward way.

Upvotes: 1

Views: 556

Answers (1)

EdChum
EdChum

Reputation: 394279

Actually this is very simple, we can call the attribute .values to return a np array of the df values, this has a method tolist(), you can assign this directly to your new column:

In [258]:
import pandas as pd
import io
t="""month  day  hour  a  b  c  d  e  f  g z
1       1     1      3   9   0   9   0   3   3     9"""
df = pd.read_csv(io.StringIO(t), sep='\s+')
df = pd.concat([df]*2, ignore_index=True)
df

Out[258]:
   month  day  hour  a  b  c  d  e  f  g  z
0      1    1     1  3  9  0  9  0  3  3  9
1      1    1     1  3  9  0  9  0  3  3  9

In [264]:
df['list'] = df[df.columns[3:]].values.tolist()
df
​
Out[264]:
   month  day  hour  a  b  c  d  e  f  g  z                      list
0      1    1     1  3  9  0  9  0  3  3  9  [3, 9, 0, 9, 0, 3, 3, 9]
1      1    1     1  3  9  0  9  0  3  3  9  [3, 9, 0, 9, 0, 3, 3, 9]

output from .values:

In [265]:
df[df.columns[3:]].values

Out[265]:
array([[3, 9, 0, 9, 0, 3, 3, 9, [3, 9, 0, 9, 0, 3, 3, 9]],
       [3, 9, 0, 9, 0, 3, 3, 9, [3, 9, 0, 9, 0, 3, 3, 9]]], dtype=object)

Upvotes: 1

Related Questions