What is the most pythonic way to combine multiple string columns to create a new Pandas series?

Question

I have a large dataframe with each row containing various amounts of text/string data (song lyrics that were webscraped and then split by line '\n'). Some columns have None values because of this. I'd like to combine all the columns that do have a value into 1 column for each row. I've attached a screenshot so you can see what I'm working with(profanity censored).

df.stack().groupby(level=0).agg(','.join) should do the trick — Umar.H, Commented Feb 27, 2020 at 20:53
Looks like you have to treat this before actually putting it in a data frame. Why do you split by lines in the first place? — rafaelc, Commented Feb 27, 2020 at 20:57
Let's not forget that being "pythonic" also requires the solution be clear and maintainable as well as concise. Exploiting the features of python is great, but if you re-read your code six months from now and wonder "it works, but how?" then you have missed the point. — tnknepp, Commented Feb 27, 2020 at 21:00

YOLO · Accepted Answer · 2020-02-27 20:55:03Z

1

To avoid NAs, here's a way using agg:

df_with_line_sentences.agg(lambda x: ' '.join(i for i in x if not pd.isna(i)), axis=1)

answered Feb 27, 2020 at 20:55

YOLO

21.8k5 gold badges24 silver badges42 bronze badges

Add a comment |

Collectives™ on Stack Overflow

What is the most pythonic way to combine multiple string columns to create a new Pandas series?

1 Answer 1

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Related