Discussion: O-scores analysis of the Billygoat is here

here
Some names marked as {new} need to be corrected to match original ones

It's a strange couple of races to try and analyze. Some of the exclamation points and question marks on the Billygoat are the direct result of following, as someone either "went for a ride", or waited for a "passenger". Then the Gruff was a case of already tired people, some of whom were putting out virtually no effort.

Something is fishy.

@J-J strange couple of races ... Some of the exclamation points and question marks .. are the direct result of ...
Great!
This is exactly why I'm doing this. Without exclamation signs you just have a time - nothing more nothing to compare with!!!!.
With o-speeds you can get any explanation you want! This is the goal- not just get time, but also compare the time with you average predicted finish time and enjoy analysis!!!!!

F.e Ross O-speed at Gruff was higher than at the main course - that means he is relatively less tired than others.

Hillary on the main course got !! well above her average apparently keeping well with Sam, who was on the expected time...
Ethan was slow on both courses ( whatever the reasons are...)
and J-J was on the expected time both times...

etc....

Enjoy

That's fine,you just have to take into account that these races have more extenuating circumstances than normal races, some having nothing to do with fitness or navigation. For example, Tracey's time was helped by the fact that Erin was running with her and waiting up when necessary, Ben was able to follow Tim, Nate was much slower than usual because he was guiding his little brother around the course, Rick stayed with Susie, etc.

(And -- ahem -- J-J was faster than expected on the Gruff. Probably because most people walked around the course.)

you just have to take into account (....)
:)
I do not have to ..., because I do not do any tweaking or anything. The number you see is your O-speed by definition involving fitness, navigation and everything else. O-speed is ratio of the time you spent on the course to the time of the average runner who would get 50 points.

You, while interpreting, have to take into account (....whatever applicable...)

Enjoy

You, while interpreting, have to take into account

That's what I meant. One has to take into account when looking at this analysis that...

! now I know where to send those who can not get it... :)

Yes, but the problem is that if you now use the data from these two runs as part of the expectations computation for future races, your base data is not reliable. For instance, Nate will now have a lower expectation than he deserves based upon his ability - albeit not based upon his performance here.

Of course, the expectation will be correct if running with his brother has made him less of an accomplished orienteer.

Perhaps these errors are small enough that they fall into the "noise region", but it seems that you could be including a data set that there is good reason to believe is 'corrupted'.

All points correct, including suggestion that everything is in the noise.

Initially, I was afraid to put this "suspicious events" data-sets into the base set and still watch carefully, but everything looks quite stable.
In practice up to date: fun of getting the score is real, while perturbations - minimal.

Here is even more questionable table:
thanks to times published in the convenient format at billygoat.org I made O-speed compilation of all Billygoats 1979-2010

Notice that average score for all events 1979-1999 is 50, because Billygoat is the only event available for these years and average runner by definition must have score 50.

Why are, say, Mike Waddington's scores so low for the 5 Billygoats that he won?

It looks to me like there's a weird effect going on, wherein people get slower as they get older, and therefore the later years get inflated in value (they must be harder, because the same people took more time to do the courses, right?). As a result, the year that I finished 5th gave me a score that's only marginally better than the three years when I led Stephen around, and this year was my best ever!

On the other hand, my times for 1994 and 1994 are completely wrong, so there's evidently some kind of data import problem (that may affect other people as well). Would you like this data as an Excel file, Shura?

Thanks for checking, J-J.

The file I got from the web automatically converted to EXCEL, so I assume my data is OK aside from human errors.
If you have data in column format, this would be great.
I had to spend a lot of time making table data into "column" data, and here are the mistakes.

Yes 1994 (and 1994??? :) ) is completely screwed up, I found an error already, it happens to be "parallel feature". Sharon was given time of Regan Paul and the rest of the list is distorted for this year.

Effect of "inflation at older dates" was noticed before and surely exists, but there is no way to fix it. Commonly proposed - "runners get slower" is wrong explanation. Here is the right explanation. In two words: "database size". Average runner for a year always gets 50 points. For years before 1999 only Billygoat is currently in the database, so all who took part in it on average got 50 points.
Anybody's score is relative to this (relatively high standard) 50 points.
Starting 2000 and up to about 2005, number of people grows and average runner is getting slower, so even though people are getting older, their scores are getting better. Starting 2005, number of people in the database is about 4000+ and average runner is quite stable.
Currently average runner at the Goat is about 85-90 points - , assuming population is no faster or slower - we have to conclude that inflation is almost two-fold.

This also answers to @BorisGr question about Mike's speed.

OK, I understand now. Yes, part of the problem is that the Billygoat attracts a higher caliber field than the average of all orienteers, so when you force the average Billygoat orieneer from the old days to have the same score as the average orienteer from now, that will bias things.

Spreadsheet on its way to the address you have listed here.

Ken Walker Jr (M32) != Ken Walker Sr (M61)

Why are some people split into two parts (e.g. Joe Brautigam/Joe double Brautigam, Alexandra Jospe/Alexandra double Jospe)? There are also some people (Nick double Duca, Martin double Hawkes-Teeter) who have "double" in their names, but no second entry.

@ ken: Ken Walker Jr (M32) != Ken Walker Sr (M61)
I know, tell me where I'm wrong and I will fix.

Sometime it is just Ken Walker, and then I have no way to find out who it is.. but now as we are improving feedback we can fix all errors.

@JJ: Doubles are there as intermediate step. First, Joe and Joseph, Alex and Alexandra are there as different people. Then I find them and rename, giving middle name "double" as a red flag. At the checkpoint, after bunch of corrections is done, I regenerate database replacing all doubles with their "identical singles". So no worry about doubles - they are error already cought but not fixed yet.
But if you See Jeff/Jeffery Charlie/Charles, Shapiro/Schapiro etc with different spelling - it needs to be fixed.
Thanks for cooperation

See http://www.billygoat.org/statistics/times.html

JJ has already done the work of matching names, spellings, etc.

am I missing some results? says I ran 1 billygoat.

JJ has already done the work of matching names
:)
yes, but only if the same job with the same names were done by all other event organizers where all this people ever ran.... When in better times we will have one database of names available for everybody to register - there will be no problem with names and they will be truly matched...

@ Joe, says I ran 1 billygoat
Most probably Joseph ran one and Joe did the rest.. up to total 19 recorded.. :)

I just regenerated the database, so all doubles should be gone and all errors detected so far are fixed ( except for 1993/1994 results which need to be redone)

Joe/Joseph Mokszycki
Krum Sergiev/Serguiev (can't keep track of how to spell his own name)
Alar Ruutopold/Ruutop?ld

Looking at some of the ordering does point out what a bizarre notion it is to try and assign a single number to describe the ability of a person considered over a 30-year span. There are people in there who advanced from being little children to top competitors, who declined from top competitors to doddering old fogeys, and others who participated only in their prime. The resulting numbers are basically meaningless. It's difficult to see the point of looking at race data over a span of more than a year for this kind of purpose.

Names queued

I agree about lack of sense for the power number over so many years. But I look at these numbers for last couple days first time, same as you, so to make decision of having sense I needed to do it... Now I think, maybe I should split them to 5 year long periods?

Actually power is calculated for every year and scores also for every year. It is just difficult to display it. So scores as they are displayed have sense. Power doesn't have sense in this table as it is only showing power at last event. But if you go to single event view, you will see correct power for this particular year
BG 2005
2007
etc.