Remove duplicates from nested list based on a string and a value

You can use itertools.groupby for grouping your elements by first index and max function with a proper key to select the max based on second element :

>>> from itertools import groupby
>>> l=[['john', 14, 'USA'], ['john', 27, 'USA'], ['paul', 17, 'USA'], ['paul', 36, 'USA']]
>>> [max(g ,key=lambda x:x[1]) for _,g in groupby(sorted(l),lambda x: x[0])]
[['john', 27, 'USA'], ['paul', 36, 'USA']]

Or as a more efficient way you can use operators.itemgetter() instead lambda :

>>> from operators import itemgetter
>>> [max(g ,key=itemgetter(1)) for _,g in groupby(sorted(l),itemgetter(0))]
[['john', 27, 'USA'], ['paul', 36, 'USA']]

I like Kasra's solution, but jsut to give another way to do it:

from collections import defaultdict

l=[['john', 14, 'USA'], ['john', 27, 'USA'], ['paul', 17, 'USA'], ['paul', 36, 'USA']]
key=defaultdict(list)
for n,a,c in l:
    key[(n,c)].append(a)
f_list = [[k[0],max(la),k[1]] for k,la in key.iteritems()]

trying my hand at incomprehensible level pythonic.

using list and dictionary comprehensions i sort, merge, and reformat

a = [['john', 14, 'USA'],['john', 27, 'USA'],['paul', 17, 'USA'],['paul', 36, 'USA']]

b = sorted(a, key=lambda x: x[0])
c = { x[0] : x[1:len(x)] for x in b }

result = [[n] + c[n] for n in c]

You can use an OrderedDict and replace the value if we find a sublist with the same name with a larger second subelement:

l = [['john', 14, 'USA'],['john', 27, 'USA'],['paul', 17, 'USA'],['paul', 36, 'USA']]

from collections import OrderedDict
d = OrderedDict()

for sub in l:
    name = sub[0]
    if name in d:
        if sub[1] > d[name][1]:
            d[name] = sub
    else:
        d[name] = sub
print(list(d.values()))

[['john', 27, 'USA'], ['paul', 36, 'USA']]

This is O(n) as it does not have to sort the list which is n log n so this will scale better than any method using sorted.

If order does not matter a normal dict will be fine:

d = {}
for sub in l:
    name = sub[0]
    if name in d:
        if sub[1] > d[name][1]:
            d[name] = sub
    else:
        d[name] = sub
print(d.values())

If you were going to sort using operator.itemgetter would be more efficient:

from operator import  itemgetter    
sorted(l,key=itemgetter(1))

If you wanted to sort the original list:

l.sort(key=itemgetter(1))