This problem is part of our Ada Lovelace Day celebration. It is for practice only and will NOT count towards your grade!
(just have fun and use it to improve your skills: there is no submission deadline!)

In this problem you should submit a function as described. Inside the function do not print anything that was not asked!

The name of the .py source file you submit to Mooshak should NOT start with a digit (you will get an error if you submit it like that).

[ALD05] Joy and the Biased AI

Joy Buolamwini is a computer scientist known for exposing bias in artificial intelligence. She is also the founder of the Algorithmic Justice League, an organization that studies and challenges harmful uses of automated systems.

While working at MIT, Joy noticed that face-detection software struggled to detect her darker-skinned face. However, the software detected a white mask that she wore...

Joy is testing a face-detection system under different conditions. She needs your help to identify the people whose faces the system repeatedly fails to detect.

The Problem

Implement the function find_failures(tests, threshold), which takes a a list of lists (matrix tests) and an integer threshold, and returns a list of people cases where a facial recognition system fails.

The details of the problem are as follows:

• tests is a list of lists (matrix) containing integers 0 and 1.
• Each row represents one test to the system and each column represents one person.
• Many tests can be performed for the same person. The same column refers to the same person in every test.
• 1 means the system successfully detected that person's face (success).
• 0 means the system failed to detect that person's face (failure).
• threshold is the minimum number of failed detections required for a person to appear in the result.
• People are numbered from 1 to n following the columns from left to right (for n columns).

For example:

tests = [
    [1, 0, 1, 0, 0],
    [1, 0, 0, 0, 1],
    [0, 1, 1, 0, 0],
    [1, 0, 1, 0, 0]
]

• 4 tests, performed on 5 different people.
• For person 1 (first column), there are 3 sucessful tests (there are three 1's) and 1 failure (there is one 0).
• Since there is only 1 failure, Person 1 would be "flagged" or "tagged" for a threshold of 1 (at least 1 failure), but not for a threshold of 2.

The function must return a list with the numbers of the people whose number of failures is greater than or equal to threshold ("flagged" people).

People must be ordered by failure count, from highest to lowest:

• If two people have equal failure counts, the person with the smaller case number comes first.
• If nobody reaches the threshold, the function should return [] (an empty list).

Constraints

The following limits are guaranteed in all the test cases that will be given to your program:

Let t be the number of tests and p be the number of people, used in tests (a matrix of t rows and p columns):

tests = [
    [1, 0, 1, 0, 0],
    [1, 0, 0, 0, 1],
    [0, 1, 1, 0, 0],
    [1, 0, 1, 0, 0]
]
print(find_failures(tests, 2))

[4, 2, 5]

• Person 1 has 1 failure
• Person 2 has 3 failures
• Person 3 has 1 failure
• Person 4 has 4 failures
• Person 5 has 3 failures

• Person 4 comes first with 4 failures.
• People 2 and 5 both have 3 failures, so person 2 comes first (because it "appears" first in the matrix, its case number (2) is lower than case number 5).
• People 1 and 3 are excluded, because they do not meet the minimum required number of 2 failures.

tests = [
    [0, 1, 0],
    [1, 0, 0],
    [0, 1, 1]
]
print(find_failures(tests, 2))
print(find_failures(tests, 3))

[1, 3]
[]

People 1 and 3 qualify when the threshold is 2, and nobody qualifies when it is 3.

tests = [
    [0, 0, 0],
    [0, 0, 0]
]
print(find_failures(tests, 2))

[1, 2, 3]

Everyone has two failures, so person numbers determine the order.