Select columns from a data frame using their index

Asked Mar 18 '20 at 15:26

Active Mar 18 '20 at 15:32

Viewed 48 times

I am using this code to get the difference between the row values for all column combinations (tups).

new_df = df[['seg_wd','ter_wd','qua_wd','qui_wd','sex_wd']]

def all_subsets(ss):
    return chain(*map(lambda x: combinations(ss, x), range(1, len(ss)+1)))

--get all combination

tups = list(all_subsets(new_df.columns))
tups = tups[5:15]

--for each combination multiple values

df1 = pd.concat([new_df.loc[:,c].diff(axis=1) for c in tups], axis=1)

However, I am getting 20 columns on the output dataframe instead of 10. Since I have repeated column names, I am unable to select the columns I want by their name. How can I select columns using their index? (Note: In this case, I want columns [1,3,5,7,9,11,13,15,17,19])

Or does anyone know a more efficient way to get the output I need?

Thanks a lot :)

edited Mar 18 '20 at 15:32

asked Mar 18 '20 at 15:26

Lcs

Does this answer your question? [Selecting pandas column by location](https://stackoverflow.com/questions/14941097/selecting-pandas-column-by-location) – mrbTT Mar 18 '20 at 15:30
No, because I need separated columns... In this case, I want columns [1,3,5,7,9,11,13,15,17,19] – Lcs Mar 18 '20 at 15:34
You can add a list to the iloc. Try `df1.iloc[:, [1,3,4,7,9,11,13,15,17,19]]` – mrbTT Mar 18 '20 at 15:38
Don't. Just don't. Having repeated column names is really a terrible idea. You should considere renaming duplicates or using a MultiIndex. – Serge Ballesta Mar 18 '20 at 15:56
@mrbTT Perfect, thank you! – Lcs Mar 18 '20 at 16:22
Serge Ballesta, thank you for the advice! – Lcs Mar 18 '20 at 16:23

Select columns from a data frame using their index

0 Answers0