1
$\begingroup$

So I plan on making a mobile app that will let students predict their final grades based on their mock exam results. I can train my model with previous years results.

X: 5 mock results

Y: Final grade obtained

However, I have the issue that sometimes, or most the times, the user may be using the app whilst not having taken ALL the mock exams yet, they may want to see if they are on track and use it once they have done maybe 3 mocks. How should I go about this for choosing my supervised algorithm/model?

$\endgroup$
0
$\begingroup$

First you need to figure out what type of variable the response is. Is it categorical (eg. A, B, C, etc) or continuous (75%, 83%, 92%, etc)? If it's categorical, it'll definitely be ordinal. This means you can explore methods like ordered logistic regression. If it's continuous, look at something like beta regression.

For the app - if the student only keys in 3 of the 5 mock exam results, you could just get them to 'guess' their results for the other 2 mock exams. In my opinion, this is better than inputing the remaining 2 mock exam outcomes. This is since the student should have a good handle on they're going to perform.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.