Comparison of values in list of tuples in python 3.x

In your example a list of tuples is being sorted. In current logic you are using the > operator to compare two tuples. This makes sense since a tuple is a string, and conforms - Python documentation :

  

Sequences are compared lexicographically by comparing matching elements. Collections that support order comparisons are sorted according to their first unequal elements (for example, [1,2,x] <= [1,2,y] has the same value as x <= y ).

Knowing this, the trick to sort a list of tuples correctly is to rearrange the elements of each tuple according to your comparison criteria. In this case, you want to transform a tuple (nome, idade, media) into a tuple (media, idade, nome ) for comparison purposes.

Assuming that the averages should be sorted in descending order and the other elements in ascending order the implementation would look like this:

def rearranjar_tupla(tupla):
    return (-tupla[2], tupla[1], tupla[0])

def ordenar(valores):
    tamanho = len(valores) - 1
    troquei = True
    while troquei:
       troquei = False
       for i in range(tamanho):
          if rearranjar_tupla(valores[i]) > rearranjar_tupla(valores[i + 1]):
            trocar(valores, i, i + 1)
            troquei = True
    tamanho -= 1
    return valores 

See working on Ideone

If averages must also be sorted in ascending order, simply reverse the tuple positions with extended slice :

def rearranjar_tupla(tupla):
    return tupla[::-1]

See working on Ideone

It is important to note, however, that in addition to learning purposes, reinventing the wheel is often not a good idea. In addition to the extra code, this basic sorting algorithm (Bubble Sort) is expensive from the point of view of Asymptotic Complexity ( O (n²) ) and very poor from the point of view cache hits and memory management (multiple copies and unnecessary swaps put pressure on the Garbage Collector ).

The Python standard API ordering engines rely on a fairly robust algorithm called