I think he's actually the only auto-qualified champion. :-) I think the rule came in for GCJ 2014 or 2015 that the winner would win a trip to the next year's finals, and Gennady won four championships starting in 2014!
On Fri, Apr 27, 2018 at 11:07 AM 'Pablo Heiber' via Google Code Jam < [email protected]> wrote: > > Your model will also learn that if Gennady solves a problem that's not a > very useful signal about the problem's difficulty. :) > > This is actually not true. Someone solving a problem is not a useful > signal if they either solve none, solve all of them or are extremely > inconsistent in what they solve. Gennady is neither of the first two > (solved some problems, but not all of them), and the third one can't > realistically happen given that you have to solve problems of a certain > level before you can tackle problems of significantly higher level in Code > Jam because of the advancement structure. > > That being said, for the years champions were auto-qualified, their > performance in online rounds is likely biased by that fact, so you may want > to exclude them. Gennady was probably an auto-qualified champion at some > point. > > Best, > Pablo > > On Fri, Apr 27, 2018 at 9:58 AM Igor Naverniouk <[email protected]> > wrote: > >> One good way to think about this is to rank the problems by the >> probability that you will solve them. >> >> Of course, you can't compute this exactly, but you can approximate it in >> many ways -- from simple averages to complex machine learning models. >> Imagine that each problem has a hidden difficulty rating, and each person >> has a hidden skill level. The probability that person X solves problem Y is >> a function of the skill of X and the difficulty of Y. If you have the full >> history of who solved which problems, you can start by assigning a skill >> rating to each person and then using those to estimate the difficulty >> values of the problems. >> >> And then there are lots of ways of iterating on that model. For example, >> you can treat it as a classic supervised learning problem -- given the >> history of previous rounds, predict who will solve which problem in the >> next round. >> >> One cool outcome of this is that you might discover patterns of problem >> types and solvers. Maybe you are really good at dynamic programming, and >> for you those problems are easier. Your model will also learn that if >> Gennady solves a problem that's not a very useful signal about the >> problem's difficulty. :) >> >> On Fri, Apr 27, 2018 at 2:08 AM, Anil M <[email protected]> wrote: >> >>> I tried to use fraction of people solved it, but it brought so many >>> Round 2, Round 3, World Finals problems up, I can't make adjustment for >>> Round because some problems in Round 2 are in fact easy if have been solved >>> by 95% of the participants(especially for small input). So I stuck with >>> absolute value. >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "Google Code Jam" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To post to this group, send email to [email protected]. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/google-code/70445ecd-35a5-4845-8cd6-2a21193f0a6e%40googlegroups.com >>> . >>> For more options, visit https://groups.google.com/d/optout. >>> >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Google Code Jam" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To post to this group, send email to [email protected]. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/google-code/CACLQfk77W5_4tAPh69ExDFnmqpS958gGCjANxyrwg4oerVQSuA%40mail.gmail.com >> <https://groups.google.com/d/msgid/google-code/CACLQfk77W5_4tAPh69ExDFnmqpS958gGCjANxyrwg4oerVQSuA%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> For more options, visit https://groups.google.com/d/optout. >> > -- > You received this message because you are subscribed to the Google Groups > "Google Code Jam" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/google-code/CANa5jcCj_okrEvB64C__f13nhTd-U0ke-F%3DCA-oiXNJc3UK6ww%40mail.gmail.com > <https://groups.google.com/d/msgid/google-code/CANa5jcCj_okrEvB64C__f13nhTd-U0ke-F%3DCA-oiXNJc3UK6ww%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "Google Code Jam" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/google-code/CAHaiWHPmDus86ar3crqZJaGjjKWvo%3Dtt3heCXr1_acB_9QMZrg%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
