Python Machine Learning to Predict Multiple Choice Quiz Template

Asked

Viewed 186 times

3

I got the templates from the last five contests at an exam stand. Templates of multiple choice (A,B,C,D,E). Let’s imagine that the test only has 5 questions, whose templates are:

          q1 q2 q3 q4 q5
Prova 1- (A  B  B  C  D)
Prova 2- (B  D  E  E  A)
Prova 3- (E  E  B  C  C)
Prova 4- (D  A  B  C  E)
Prova 5- (C  C  A  B  D)

I want to know the likelihood, in the next contest, q3 was E, and the feedback of the next q1 was A, of q2 was B, q4 was E and q5 was D:

 q1  q2 q3 q4 q5
(A   B  ?  E  D)

I couldn’t find anything like this on forums in Brazil and abroad. Which classifier should I use in Python? Can someone help?

python machine-learning

  • You can start your analysis by observing the entropy of data.

  • Like everything in machine Learning, the answer is: depends, you can use a number of different classifiers depending on how you decide to address the problem, my first choice would be collaborative filtering (https://en.wikipedia.org/wiki/Collaborative_filtering) Note that the animation is very similar to what you want to do. But note that if the alternatives are random between yes, machine Learning can not do much, as the friend above suggested, start by looking at the entropy of the data to know whether or not there is some kind of pattern.

1 answer

1

Without knowing what the PRNG used by the bank to generate evidence, it will be very difficult to make this kind of forecast.

In short, by discovering the PRNG used and its respective seed, you can run the algorithm on your machine and, consequently, determine the order of the chosen responses.

I don’t think machine Learning will be very useful in your case because the distribution of numbers of modern Prngs is very good, that is, your machine Learning algorithm simply will not be able to identify a pattern and will only be kicking all the time.

This is all I’m saying is based on the assumption that the bank used a computer to scramble the answers. If a human has done the shuffling, machine Learning may be a viable option, but:

  1. You will need a lot of data generated by the same human
  2. Most likely it wasn’t a human

Browser other questions tagged python machine-learning

You are not signed in. Login or sign up in order to post.