根据最大值删除元组列表中的重复项

Question

Suppose I have a list of tuples like this:

[('Machine1', 88), ('Machine2', 90), ('Machine3', 78), ('Machine1', 90), ('Machine3', 95)]

And I want to filter the list such that I only have the highest values pertaining to each tuple pairing. So in this example the filtered list would be:

[('Machine2', 90),('Machine1', 90), ('Machine3', 95)]

I basically want to remove duplicates by the highest value. I know set only removes exact duplicates so I won't be able to do that here. I thought another method I could use would be to use a dictionary and update it while iterating through the list if a higher value was seen. However, what is a more pythonic way to approach this?

Answer 1

One solution with simple dict

d = {}
for machine, value in l:
    d[machine] = max(d.get(machine, -float('inf')), value)
print(list(d.items()))

Outputs

[('Machine1', 90), ('Machine2', 90), ('Machine3', 95)]

---

Using pandas (for fun)

>>> pd.DataFrame(l).groupby(0).max().to_dict()[1].items()
[('Machine1', 90), ('Machine2', 90), ('Machine3', 95)]

Answer 2

Here's one solution using collections.defaultdict . The idea is to iterate your list of tuples and append to lists. Then use zip with map + max to create the desired result.

from collections import defaultdict

L = [('Machine1', 88), ('Machine2', 90), ('Machine3', 78),
     ('Machine1', 90), ('Machine3', 95)]

d = defaultdict(list)

for name, num in L:
    d[name].append(num)

res =  list(zip(d, map(max, d.values())))

Result

[('Machine1', 90), ('Machine2', 90), ('Machine3', 95)]

Answer 3

It may be possible to use the groupby operator in itertools:

>>> import itertools as it
>>> [ (k, max( list(zip(*g))[1])   ) for k,g in it.groupby(sorted(data), key=lambda m: m[0])]

Remember that the data is sorted, so you could also do:

>>> [ (k, list(zip(*g))[1][-1]   ) for k,g in it.groupby(sorted(data), key=lambda m: m[0])]