Python sorting multidimensional dict by a specific column

Question

I know there are tons of questions about this but I'm trying to sort the dict below by the hitrate column.

data = {
    'a': {'get': 1, 'hitrate': 1, 'set': 1},
    'b': {'get': 4, 'hitrate': 20, 'set': 5},
    'c': {'get': 3, 'hitrate': 4, 'set': 3}
}

I tried a bunch of things, the most promising being the method below which seems to error out.

s = sorted(data, key=lambda x: int(x['hitrate']))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 1, in <lambda>
TypeError: string indices must be integers, not str

Can I get some help with this please?

Thanks!

Answer 1

Iterating a dict yields the keys, so you need to lookup x in the dict again:

sorted(data, key=lambda x: int(data[x]['hitrate']))

If you want the values too, then sort the items:

sorted(data.items(), key=lambda item: int(item[1]['hitrate']))

Answer 2

Using a dict as an iterable will only cause the keys to be iterated and not the values, so x in your lambda will only be "a", "b", and "c" You're basically doing "a"["hitrate"], which causes a TypeError. Try using x as keys into your dictionary.

>>> data = {
...     'a': {'get': 1, 'hitrate': 1, 'set': 1},
...     'b': {'get': 4, 'hitrate': 20, 'set': 5},
...     'c': {'get': 3, 'hitrate': 4, 'set': 3}
... }
>>> s = sorted(data, key=lambda x: int(data[x]['hitrate']))
>>> s
['a', 'c', 'b']