Pandas multiindex dataframe: 2 values on same level

Question

I want to create a multiindex Dataframe in Pandas, however, I have 2 values (string and float), that are unique and to my understanding should be on the highest level. Each of my data sets has one string with a respective (float) value and 4 features that have their respective values in 16 columns.

What is the correct/pythonic way to create such a dataframe?

1.) Having the value as a level?

2.) Inserting the value 4 times in all columns of the set as its own column?

3.) Something more elegant I am not aware of?

If you think this question is for whatever reasons inappropriate, please let me know in a short comment why and don't just down rate. Thanks a lot!

"I have 2 values ( string and float)" dont forget to add a sample of them here — Bharath M Shetty
– Bharath M Shetty, Commented Oct 27, 2017 at 11:31
@Bharath shetty, jezrael : added sample data as a picture, didn't know how else to do it — MaxS
– MaxS, Commented Oct 27, 2017 at 11:46

jezrael · Accepted Answer · 2017-10-27 12:00:50Z

1

I think the best is create DataFrame with:

Name, Features, value <-MultiIndex with 3 levels
Pos1 - Pos16 - columns

But all depends of what you need do later with data.

answered Oct 27, 2017 at 12:00

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

MaxS Over a year ago

so what you would suggest is option 1 in the question, right? I want to frequently use the values and correlate them with values of the features and maybe also use them as a variable to plot my data

jezrael Over a year ago

Yes, but in my opinion the best is dont mix string columns with numeric. Because if want some easy operiation like df = df * 10, then if string column it fail. But if Strings columns are in MultiIndex and all data columns are numeric, it working nice. The best is distinguish data and metadata (in columns, index)

MaxS Over a year ago

following your argument it would be best to put the strings in the index? I thought the first level of the multiindex df is treated like some kind of index. Is this incorrect?

jezrael Over a year ago

It is still up tou you ;) If it is index or columns, it depends for your using. But simpliest is selecting by columns like df['col']. As MultiIndex column like df.xs('col', axis=1, level=0). But all works ;)

MaxS Over a year ago

Thanks for that last comment, it really made things clear for me!

Collectives™ on Stack Overflow

Pandas multiindex dataframe: 2 values on same level

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related