Problem category prediction part in termicoder
This aims at prediction of problem category using AI (ML) using huge amount of questions available online which have already got tags. Though achieving high accuracy may be a problem but it can be a quite an exciting problem in terms of the result. I expect that various exciting relations can be found with following Factors :-
- Number of submissions
- Rating of the author and his existing questions
- NLP on Problem statement
- Test Cases
- Execution time and memory used by the existing solutions as well as time taken for first submission to appear
- Time Limit , Memory Limit given
- Constrains on test data
- Distributions of submission with language (ex- Big Integers are difficult to code in c/c++ etc)