Statistics 2

[Tom Grahame]

Here is my second pass at modelling statistics:

https://gist.github.com/tfgrahame/7603346

I tried to go through all of the contributions made to the list but stopped at American Football - I still have some outstanding questions and there were just too many to deal with!

Some statistics I did not understand so please don't be offended if they did not make it into this cut.

I tried to adopt the class hierarchy design pattern that schema.org uses.

I took a gamble and created Motorsport statistic

Some questions:

Is there merit in splitting out at a high level, Team and Individual stats?

How far do we go with the Golf example where we split stats out by group and type?

For Gymnastics, is there any need to split between Team and Individual when they share the same properties?

Even across a few sports there is a lot to deal with. I recommend that we cut this down before making a proposal - perhaps just to the absolute minimum generic statistics that we can apply to either teams or individuals.

Best,

Tom

[Vicki Tardif Holland]

I think for the first version, we may want to limit ourselves to high level, common statistics. The individual sports get messy really fast, and will probably need specific subclasses.

To that end, I would not go really deep into golf right now. If we can get a basic sense of a player's score for a given tournament and the rankings, we get a long way to modeling most sites.

- Vicki

Vicki Tardif Holland | Ontologist | vta...@google.com 

 

--
You received this message because you are subscribed to the Google Groups "sports-schema-collab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sports-schema-co...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Tom Grahame]

Here's another sample:

https://gist.github.com/tfgrahame/7638541

This is as general as I could manage and even now it stops making sense across sports. There's no strict notion of draws and losses in Motorsport for example.

The only values that could make sense in a cross sport context are really points and rankedPosition and even then I suspect that things don't always make sense.

At this stage we have a wide variety of properties to choose from. I suggest we try and achieve a consensus as to which really do apply in a cross sport context and therefore make the cut into an initial proposal.

Regards,

Tom

[Vicki Tardif Holland]

Sports are not homogeneous enough to come up with a truly universal set of properties. Wins, losses, draws, and ranked position are common enough that they should be included even if some sports do not use every property. 

I am on the fence about the various points properties, but I suspect they are important for sports common outside of North America.

Vicki

Vicki Tardif Holland | Ontologist | vta...@google.com 

 

[Tom Grahame]

I think this is the decision I am seeking clarity on:

they should be included even if some sports do not use every property. 

Thanks,

Tom 

[Paul Kelly]

Just an update to apologize for not getting that core-stats elaboration done. Had some fires to put out at work. Hope to get to it on weekend or Monday. Have a good weekend everyone!

[Paul Kelly]

Please let me know if this is OK:

https://gist.github.com/polvough/7881271

I tried to get my parentheses and brackets in different colours but couldn't figure that out.

[Jason Johnson]

Thanks for putting this together Paul.  Looks like a solid start and gives us the mapping we need to understand their application across sports.

>> To unsubscribe from this group and stop receiving emails from it, send an email to sports-schema-collab+unsub...@googlegroups.com.

[Paul Kelly]

As promised another draft:

https://gist.github.com/polvough/7915545

I hope this isn't too team-sports now? I checked on golf, tennis, and NASCAR and I think we're covered. I put bonusPoints back in because of NASCAR.

Includes the following notes:

1) pointsDifference, pointsScoredAgainst, pointsScoredFor changed to make clear they refer to actual event scores
2) pointsTotal and scoreTotal removed. Tom please set me straight here. What do they mean as opposed to score and points?
3) rankedPosition is now just rank
4) pointsDeducted and lossesOvertime mainly apply to a single sport but are core stats in the standings for soccer and ice hockey respectively. Leaving them out might raise problems.
5) streak changed to win-streak and loss-streak

I'd like to hear from Tom next week (he's on vacation) before we finalize.

> >> To unsubscribe from this group and stop receiving emails from it, send an email to sports-schema-co...@googlegroups.com.

> >> For more options, visit https://groups.google.com/groups/opt_out.
> >
>
>

> --
> You received this message because you are subscribed to the Google Groups "sports-schema-collab" group.

> To unsubscribe from this group and stop receiving emails from it, send an email to sports-schema-co...@googlegroups.com.

[Tom Grahame]

This looks good to me and I now see what Jason needed with regards to the clarification of terms.

pointsTotal and scoreTotal can be removed. There was a scenario concerning the simultaneous rendering of the points of a single game and the total points accumulated for a league but I think that is an edge case that can be ignored.

Minor syntactic point: we should probably be using winStreak and lossStreak.

Tom

[Paul Kelly]

On 2013-12-13, at 7:58 AM, Tom Grahame <tfgr...@gmail.com> wrote:
> Minor syntactic point: we should probably be using winStreak and lossStreak.

I've made that correction.