When using Pandas DataFrame.sort(), can I make it actually renumber the rows?

Question

I am always surprised by this:

> data = DataFrame({'x':[1, 2], 'y':[2, 1]})
> data = data.sort('y')
> data
   x  y
1  2  1
0  1  2

> data['x'][0]
1

Is there a way I can cause the indices to be reassigned to fit the new ordering?

Answer 1

For my part, I'm glad that sort doesn't throw away the index information. If it did, there wouldn't be much point to having an index in the first place, as opposed to another column.

If you want to reset the index to a range, you could:

>>> data
   x  y
1  2  1
0  1  2
>>> data.reset_index(drop=True)
   x  y
0  2  1
1  1  2

Where you could reassign or use inplace=True as you liked. If instead the real issue is that you want to access by position independent of index, you could use iloc:

>>> data['x']
1    2
0    1
Name: x, dtype: int64
>>> data['x'][0]
1
>>> data['x'].iloc[0]
2