How are the problems in Problem archive classified

Feel free to ask any questions related to the COJ.
Post Reply
Looney
Posts: 5
Joined: 3 years ago
Gender: None specified

How are the problems in Problem archive classified

Post by Looney » 3 years ago

Hi All,
I am curious to find out how does difficulty level vary between Level 1 to Level 5 problems. I have been using the percentage of accepted solutions as a rough metric, but i am really curious if more is known about this.
What i am looking for is some guidelines as, in Level 1 problems can brute forced and level 5 problems require modified versions of some latest research algorithms etc.
Also what metric is coj using to classify problems in levels ?
Thank you in advance.



humbertodiaz
Posts: 97
Joined: 4 years ago
Gender: None specified

Re: How are the problems in Problem archive classified

Post by humbertodiaz » 3 years ago

As far as I know, Problemsetters provide the difficulty values when submitting an exercise. That conclusion is based on having submitted some exercises for COJ a few months ago. I was asked to provide that information. In fact, there is a difficulty level associated with each category - a problem may be Geometry (2), Data Structures (3), for instance.

Unfortunately, this has lead me to question the validity of the measurement. It's subjective and experience-based. Lower-level users would perceive higher-level exercises as being really hard without much distinction. Higher-level users may perceive things that they are comfortable with as easy, and things that they don't know as well as harder. Case in point, I'm sure some users around here find various string problems to be straightforward. Despite my score, those kinds of problems are currently out of my reach, and it shows in my stats. Perhaps a point-based system associated with the required algorithms would be a better metric.

Acceptance rates, by the way, are not quite a good metric either. The reason being that sometimes an exercise is clearly insurmountably hard for lower-level users and they don't try it. Meanwhile, a higher-level user with the benefit of experience could solve the exercise and would be far less likely to make mistakes. This can lead to nasty situations when problems are chosen for contests based on acceptance rates - I saw this firsthand once. XD

Post Reply

Return to “FAQ”