willyg23
commented
7 months ago Do American fighters win more or less often when fighting outside of the US? Do non-American fighters win more or less in the US? In their home country? Or perhaps a more relevant question that encapsulates both questions: Do fighters win more or less in their home country? And vice versa not in their home country? Since fighters fight in the US so often though, I wouldn't be surprised if non-American fighters do just fine in the US. So maybe make an additional one that counts any country that isn't their home country, except the US.
Are there more knockouts, subs, decisions, etc. in certain cities?
Do fighters with with or without nicknames win more or less compared to each other? Do they fight differently?
- these would be done with the new data set
on your old project, add a visualization of the predictions
of champions who go on a losing streak (or have a lot of losses, "fall from grace') after losing the belt, what do they have in common?
of underdogs who win (lets say +200 or bigger) are there any trends they have in common? Win/loss streak? Amount of wins as the underdog? (Rose Namajunas for example) Physical attributes like height, reach, etc.? Weight class? (probably heavyweight, since heavyweight fights can end the quickest)? Lots of wins via a certain method? Age of them or their opponent? Age/Physical attribute differentials?
Same question for favorites who lose
do one specifically on seeing height and reach differentials' correlation to winning. Big height diff + big reach diff, small/equal height diff + small/equal reach diff, Big height diff + small reach diff, small height diff + big reach diff,
of unlikely fight outcomes that happen i.e (+1600 sub for Jiri when he fought glover texeria), do they have anything in common per outcome type(sub,ko/tko, Split-dec, ...)
-Fight prediction model. of course - use the old data set here. not enough stats in the new data set. -Does height give an advantage to win a fight? Does any specific stat (strike landed, head/body/leg strikes, takedown defense, etc.) increase or decrease with it? Same questions also can be applied to reach. - could use either data set. The old data set has readily available average statistics for fighters. However,. fighters don't have a unique id, like they do in the new data set. So we wouldn't have to make a way to store fighters with the new data set, which would be nice. -Are there any trends we see with certain referees? Do some referee's fights end in knockouts or submissions more than others? - new data set, old data set doesn't have any mention of referees -Prediction of future fights - This might be easier to do on the new data set as well, with fighters having fighter ids. Then you could just plug in one fighter ID against another. From a UI perspective, you'd want users to be able to type in a fighters name in one box, and another's in another box, and see the predicted fight outcome. It'd be cool to see really in-depth predictions too, in addition to who's goin to win. Round and time of the finish (if applicable), method of finish(KO, SUB, Doctor stoppage), method of the method of finish (KO->kick,knee,punches,elbows SUB->rear naked choke, armbar, kimura, etc.), how many knockdowns there are, and by which fighter(s), -Predict all UFC 300 fights!!