Kaushik Kaushik - 2 months ago 5x
Python Question

what is the difference between x[1,2] and x[1][2] in hierarchy indexing for series in python?

I have a series


that looks like this:

1 a -0.068167
b -1.036551
c -0.246619
d 1.318381
2 a -0.119061
b 0.249653
c 0.819153
d 1.334510
3 a 0.029305
b -0.879798
c 1.081574
d -1.590322
4 a 0.620149
b -2.197523
c 0.927573
d -0.274370
dtype: float64

What is the difference between x[1,'a'] and x[1]['a']. It gives me the same answer. I am confused as to what the difference internally means? When should I use the above two indexes?


This explanation is from the numpy docs, however I believe a similar thing is happening in pandas (which uses numpy inside, using "indexers" to provide a mapping between a (possibly) named index and the underlying integer-based index).

So note that x[0,2] = x[0][2] though the second case is less efficient as a new temporary array is created after the first index that is subsequently indexed by 2.

Here are the timings for your series; the first method is around 30 times faster:

In [80]: %timeit x[1, 'a']
100000 loops, best of 3: 8.46 µs per loop

In [79]: %timeit x[1]['a']
1000 loops, best of 3: 274 µs per loop