How to get a sample dataset from large datafram while keeping the proportion of categorical variables in a column

Asked Sep 22 '20 at 11:47

Active Sep 22 '20 at 11:52

Viewed 261 times

I have a large DataFrame with 2 million observations. For my further analysis, I intend to use a relatively smaller sample (around 15-20% of the original DataFrame) drawn from the original DataFrame. While sampling, I also intend to keep the proportion of categorical values present in one of the columns intact.

For eg: if one column has 5 categories as its values: red(20% of total observations), green(10%), blue(15%), white(25%), yellow(30%); I would like the column in the sample dataset to also show the same proportion of different categories.

Please assist!

edited Sep 22 '20 at 11:52

asked Sep 22 '20 at 11:47

AmanAgrwl

2

Does this answer your question? [Sample each group after pandas groupby](https://stackoverflow.com/questions/36390406/sample-each-group-after-pandas-groupby) – Michael Szczesny Sep 22 '20 at 11:56
I tried this but it turns the categories into something similar to an index and therefore aren't practically a column anymore. How can I, say, ungroup it back!? – AmanAgrwl Sep 22 '20 at 12:00
2

You can try adding the argument as_index = False in the groupby() – Rajesh Bhat Sep 22 '20 at 12:23
that works perfectly! Thanks! – AmanAgrwl Sep 22 '20 at 16:41

How to get a sample dataset from large datafram while keeping the proportion of categorical variables in a column

0 Answers0