How to change from index to multiindex - pandas

Question

I've got a data frame structured as below:

dict1 = {'id': {0: 11, 1: 12, 2: 13, 3: 14, 4: 15, 5: 16, 6: 19, 7: 18, 8: 17},
 'var1': {0: 20.272108843537413,
  1: 21.088435374149658,
  2: 20.68027210884354,
  3: 23.945578231292515,
  4: 22.857142857142854,
  5: 21.496598639455787,
  6: 39.18367346938776,
  7: 36.46258503401361,
  8: 34.965986394557824},
 'var2': {0: 27.731092436974773,
  1: 43.907563025210074,
  2: 55.67226890756303,
  3: 62.81512605042017,
  4: 71.63865546218487,
  5: 83.40336134453781,
  6: 43.48739495798319,
  7: 59.243697478991606,
  8: 67.22689075630252},
 'var3': {0: 1, 1: 1, 2: 1, 3: 1, 4: 1, 5: 1, 6: 2, 7: 2, 8: 2}}
ex = pd.DataFrame(dict1.to_dict()).set_index('id')

id is set as an index, but now I would like to create a MultiIndex from var3 and id. But my following attempt fails:

ex.set_index(['var3', 'id'])

How can I then set a MultiIndex straight from Index? I know I can reset_index first and then set a MultiIndex, but it feels there has to be more elegant way.

long way - ex.groupby('var3').apply(lambda x: x.assign())

splash58
– splash58

2019-11-01 19:38:37 +00:00
Commented Nov 1, 2019 at 19:38 — splash58
– splash58, Commented Nov 1, 2019 at 19:38

PMende · Accepted Answer · 2019-11-01 19:42:03Z

4

DataFrame.set_index has an append argument, which is False by default.

If you have a DataFrame already indexed by "id", and you'd like to append "var3" to that, simply invoke:

new_df = ex.set_index("var3", append=True)

As suggested by @piRSquared in the comments, you can also swap the order if you would like "var3" to come first by method chaining a call to swaplevel. I.e.:

new_df = ex.set_index("var3", append=True).swaplevel(0, 1)

edited Nov 1, 2019 at 19:42

answered Nov 1, 2019 at 19:35

PMende

5,4903 gold badges21 silver badges26 bronze badges

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

2 Comments

piRSquared Over a year ago

ex.set_index('var3', append=True).swaplevel(0, 1)

Anton vBR Over a year ago

@piRSquared Yeah, that would work, but in that case I'd say my solution is easier to read.

Anton vBR · Accepted Answer · 2019-11-01 19:22:52Z

4

Like this:

ex.set_index(['var3', ex.index])

answered Nov 1, 2019 at 19:22

Anton vBR

19k6 gold badges47 silver badges47 bronze badges

4 Comments

piRSquared Over a year ago

For the record. This might be more readable to some and it should be quicker.

Anton vBR Over a year ago

@piRSquared heck, even ex.reset_index().set_index(['var3', ex.index]) is more readable than your pro solution :)

piRSquared Over a year ago

I'd not say that is my recommendation. I just wanted to show that swaplevel was necessary to make the answer accurate. BTW, you could do this in place df.index = [df.var3, df.index]

Anton vBR Over a year ago

@piRSquared hehe, just kidding. Yeah well not really as var3 would still exist in the frame.

Collectives™ on Stack Overflow

How to change from index to multiindex - pandas

2 Answers 2

2 Comments

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related