CODEWITHSUNDEEP
pythondatetimepandas
user308827
user308827

Reputation: 22001

Converting multilevel pandas index to datetime

I have the foll. dataframe:

          col_a     col_b  
4 15  69.717679  0.09701  
  16  69.717679  0.09701  
  17  69.717679  0.09701  
  18  69.717679  0.09701  
  19  69.717679  0.09701

It has a multi-index with the first level representing the month(4) and the next level representing the day of the month (1..31). How do I convert the multilevel index to a datetime index? The year value is 2013.

I tried this:

pd.to_datetime(df_past.index.levels[1] + df_past.index.levels[0] + 2013, format="%d%m%Y")

but get the error:

*** ValueError: cannot evaluate a numeric op with unequal lengths

Upvotes: 1

Views: 911

Answers (2)

Alexander
Alexander

Reputation: 109636

The following list comprehension/zip method will get you the dates as datetime objects.

import datetime as dt

new_index = [dt.datetime(2013, month, day) 
             for month, day in zip(*[df.index.get_level_values(i) for i in (0, 1)])]

Upvotes: 1

joris
joris

Reputation: 139222

To get the values of a level of the index, you need .get_level_values() instead of .levels:

pd.to_datetime(2013 * 10000 + df_past.index.get_level_values(0) * 100 + 
               df_past.index.get_level_values(1), format="%Y%m%d")

Further, if you add those as integer values, you need to multiply with 10000/100 to get the correct format. An alternative is to astype(str) each of them, and then just adding (concatenating) will work.

Upvotes: 4

Related Questions