Review of Car Classification System

[James Mewett]

With the 2006 season over it is time to look at the results and see if they reveal any weaknesses in the Car Classification System. The data you entered in the Car Classification Database has been very useful in looking at the results in more detail than previously. Specifically your results (if you are a regular competitor) have been PAXed not by the class PAX factor but by one that relates specifically to your exact level of preparation. This data has been plotted against various factors (Weight, HP, SUSP and total number of PIPs claimed) in order to find if the Classification System favours particular cars (say light-weight cars).

The results are very encouraging, but you are invited to have a look at the attached analysis and draw your own conclusions.

[thgear]

great work!

have you had a chance to plot the "pax averages" for each competitor in order, against this years results?

how do they compare?

has there been discussion of perhaps using PAX to score the overall championship, and our old system for determining class winners?

[Todd #65]

Thank you James. That looks like a tonne of work.

As the information is presented, a car perfectly suited for a certain track on 1 event is skewing the charts... is it not?

I agree with Serge that the PAX averages would be interesting to compare...

[Carguy]

A job very well done James, as usual. :) :cool:

I realize a lot of work goes into these analyses and I'm not looking to heap more on you... Having said that, I believe a case might be made for or against uber tires by applying the same calculations to last year's data. Comparing the normalized results of drivers who ran non-uber tires but switched to uber tires this season could give us an idea of their effectiveness. Note that the analysis should not be limited to drivers who switched to uber tires, but all regular competitors since other conclusions such as improving driver ability could be reached.

[thgear]

you know....

there are alot of IT students that are looking work in this sort of thing, and some will do it for free just so they can put it on their resume..

we can get like 10 of these guys to crunch all the numbers we want them too.. imagine!

[Carguy]

Looking at the graphs a bit more closely - in particular Avg and Best IPAXed Results vs SUSP - make me wonder how well last year's suspension convergence worked out. The upward slope suggests that cars with a low basic suspension ratings fare slightly better than cars with higher basic suspension ratings. What might be more revealing is the same plot using actual suspension ratings. Hmmmm...

[ScotcH]

Looking at the graphs a bit more closely - in particular Avg and Best IPAXed Results vs SUSP - make me wonder how well last year's suspension convergence worked out. The upward slope suggests that cars with a low basic suspension ratings fare slightly better than cars with higher basic suspension ratings. What might be more revealing is the same plot using actual suspension ratings. Hmmmm...

Seems to me that is exactly the reason why convergence was required ... imagine the results if these cars were PIPed lower (since without conversion, their sPIPs would be worth 1 PIP instead of the 1.x with convergence, thus lowering their final PI)

[Carguy]

Seems to me that is exactly the reason why convergence was required ... imagine the results if these cars were PIPed lower (since without conversion, their sPIPs would be worth 1 PIP instead of the 1.x with convergence, thus lowering their final PI)

Agreed Arek. My question is did we go far enough?

[ScotcH]

Agreed Arek. My question is did we go far enough?

Ahhh ... my thought was the same :)

[G-ForceJunkie]

Those 4 dots on the 35% have quite a bit of leverage on that line, take those out and the line would be quite a bit straighter I would think.

[Todd #65]

Ahhh ... my thought was the same :)

I'm not sure how you are coming to these conclusions based on the data provided....

Your discussions foundation is based on a single best score of the year. Unless you look at average best score for a vehicle&driver then you are only accounting for the performance of the vehicle/driver at one specific track configuration.

Basically, this analysis consists of the one best pounding given by that car over the year. Meaning.... if we ran at DDT 8 times and then ran at Mosport 1 time. My car would show as being dominant in it's class because of 1 good score...

[Carguy]

Those 4 dots on the 35% have quite a bit of leverage on that line, take those out and the line would be quite a bit straighter I would think.

That's why I think we should be using actual suspension ratings and taking into consideration the basic suspension rating starting point (i.e. the difference between the 2 perhaps). Again, I say hmmmm... still thinking about whether this is possible or not.

[thgear]

I'm not sure how you are coming to these conclusions based on the data provided....

Your discussions foundation is based on a single best score of the year. Unless you look at average best score for a vehicle&driver then you are only accounting for the performance of the vehicle/driver at one specific track configuration.

Basically, this analysis consists of the one best pounding given by that car over the year. Meaning.... if we ran at DDT 8 times and then ran at Mosport 1 time. My car would show as being dominant in it's class because of 1 good score...

hold on

there are two lines "best" and "average"

the average would represent your overall score, and best, i think was used in James' hypothesis that any car could beat pax if it really wanted to.

[Carguy]

I'm not sure how you are coming to these conclusions based on the data provided....

Your discussions foundation is based on a single best score of the year. Unless you look at average best score for a vehicle&driver then you are only accounting for the performance of the vehicle/driver at one specific track configuration.

Basically, this analysis consists of the one best pounding given by that car over the year. Meaning.... if we ran at DDT 8 times and then ran at Mosport 1 time. My car would show as being dominant in it's class because of 1 good score...

Good point Todd, averages might be better. However, assumptions are being made on so many factors. What if a particular car does well at the Mosport big track where horsepower is rewarded on the back straight, and suspension (and braking) don't factor in as much as on a tighter track... What about the weather conditions? Yes, they would be the same for everyone but what would be the effect on individual cars? Case in point Derek's turbo Porsche at TMP in the heat was noticeably slower. Or were other factors contributing? Maybe we need to perform the analysis on a track that rewards better suspensions. :)

[thgear]

i shaved 0.7 seconds off my TMP lap by driving with my windows down (cuz one of them collapsed into the door)

go figure... somethings are just impossible to quantify..

but i think this is a step in the right direction, at the very least this analysis more or less proves that our system works!

[ScotcH]

but i think this is a step in the right direction, at the very least this analysis more or less proves that our system works!

Absolutely! I think that was the main reason for posting this ... to show everyone that fundementally, things are OK :)

[ScotcH]

I'm not sure how you are coming to these conclusions based on the data provided....

Your discussions foundation is based on a single best score of the year. Unless you look at average best score for a vehicle&driver then you are only accounting for the performance of the vehicle/driver at one specific track configuration.

Basically, this analysis consists of the one best pounding given by that car over the year. Meaning.... if we ran at DDT 8 times and then ran at Mosport 1 time. My car would show as being dominant in it's class because of 1 good score...

Well, actually I'm looking at the 35% as well as the 50% data ... both seem a bit low compared to the distribution of the other data points. As has been noted, there is a "best" and an "average" data point. Of course you're right ... it is likely due to the difference in tracks, but that can't be avoided, unless the data is replotted per track. Of course it does not help that both you guys a very good drivers. :)

[Carguy]

Absolutely! I think that was the main reason for posting this ... to show everyone that fundementally, things are OK :)


I agree Arek, but we still have to dig deeper into the results and the data. For instance the official results lists 54 competitors who completed at least 4 events, yet the analysis is based on 43 competitors one of which only ran 3 events. I'm sure there's a good explanation for this... James? :cool:

[James Mewett]

I agree Arek, but we still have to dig deeper into the results and the data. For instance the official results lists 54 competitors who completed at least 4 events, yet the analysis is based on 43 competitors one of which only ran 3 events. I'm sure there's a good explanation for this... James? :cool:

Very simple ... all the data was entered the hard way. Although the info is on the web in the classification database and in the results pages it doesn't just pop nicely into a spreadsheet without lots of massaging or simply manual entry. This is a real disincentive to including everyone, and in any case I figured the regular runners are most interesting in that they provide more data on more tracks. I think I got everyone that ran a full season and then a few more for good measure. There is no real consistency as to how I picked the few more but shince I had entered their data I did include it.

[James Mewett]

A job very well done James, as usual. :) :cool:

I realize a lot of work goes into these analyses and I'm not looking to heap more on you... Having said that, I believe a case might be made for or against uber tires by applying the same calculations to last year's data. Comparing the normalized results of drivers who ran non-uber tires but switched to uber tires this season could give us an idea of their effectiveness. Note that the analysis should not be limited to drivers who switched to uber tires, but all regular competitors since other conclusions such as improving driver ability could be reached.

If someone knew who was running Uber tires, when, and what condition they were in for each event some useful data could be extracted. I am not volunteering to try and fill in the blanks here as I see it being a monumental hand waving exercise.

[James Mewett]

great work!

have you had a chance to plot the "pax averages" for each competitor in order, against this years results?

I contemplated it but quickly gave up. I'll tell you why:

- do I use the best 6 events?
- for those that ran 5 and use organizer points for the 6th do I do the same (not really fair if you really want to know who drove best, but unfair not to since they did run a full season by our standards)
- so maybe I should use the best 5 events?

No matter what approach is used the result is contrived.

Anyway, the data is in the analysis and you are welcome to massage it as you see fit and let us know who the IPAX average winners were by whatever method you see fit!:rolleyes:

[thgear]

Anyway, the data is in the analysis and you are welcome to massage it as you see fit and let us know who the IPAX average winners were by whatever method you see fit!:rolleyes:

its in PDF tho.. haah

unless there is a way to extract the chart into an excel spreadsheet

do you have the original as an excel somewhere?

[James Mewett]

Looking at the graphs a bit more closely - in particular Avg and Best IPAXed Results vs SUSP - make me wonder how well last year's suspension convergence worked out. The upward slope suggests that cars with a low basic suspension ratings fare slightly better than cars with higher basic suspension ratings. What might be more revealing is the same plot using actual suspension ratings. Hmmmm...

The problem is that only the Presswoods show up at that end of the SUSP spectrum and they both drove very well so they bias the trend line. As always the problem is to try to figure out how much of the result is due to the classification and how much is due to the drivers. In this case I think the Presswoods deserve a fair bit of the credit.

[Chris91GT]

What if a particular car does well at the Mosport big track where horsepower is rewarded on the back straight, and suspension (and braking) don't factor in as much as on a tighter track...

Here's a curious question... now that we have data, can we correlate lap times vs HP? I think the "fact" that Mosport GP is a HP track is a myth. I don't even think it's a handling track per se. I suspect that if you correlate ET vs HP, SUSP, and shear grip that shear grip will win.

My top speed at Mosport is only 205kph or so, something that Hanif handily beats. Hanif also has a better handling car. But I suspect that my lap times would be quicker than his at Mosport. Why? Shear grip!

I'd love to see an analysis...

[James Mewett]

its in PDF tho.. haah

unless there is a way to extract the chart into an excel spreadsheet

do you have the original as an excel somewhere?

Yes, but I am not volunteering it! I want you to feel my pain. :p

You can copy and paste from a pdf you know ...

[thgear]

i'm telling you man... IT students, i'll look into it as a one time contract position (no fees) just so they can throw it on their resume

"helped compile raw data into usable to the organization information" :D

[craig]

Bah. What do IT students know about statistics and/or data analysis?

For example, I'd be willing to bet that there isn't an IT student anywhere who could explain if, how, and why ecological correlation may affect the results presented (without first going to Google :p ).

Anyhow, either you believe the system works, or you don't. Belief, not numbers. If most people believe the system works reasonably well, then it does. Otherwise, you'll venture into reductionism.

[CobraStang]

I think the "fact" that Mosport GP is a HP track is a myth. I don't even think it's a handling track per se. Its not a HP track, or a handling track. Its a "depends" tracks. Statistically, your lap time is inversely proportionate to the number of dependents you have. That's why the single guys do well at the track;they're not worried about leaving a grieving family behind!



OK, I made that up. Sort of.

[G-ForceJunkie]

Its a "depends" tracks.
LOL! ;)

[thgear]

Its not a HP track, or a handling track. Its a "depends" tracks. Statistically, your lap time is inversely proportionate to the number of dependents you have. That's why the single guys do well at the track;they're not worried about leaving a grieving family behind!



OK, I made that up. Sort of.

also i guess "depends if you have a car or not" ;)

[Slowpoke]

And the only guys with enough money to race on this track regularly are wearing "Depends". ;)

[h-bomb]

My top speed at Mosport is only 205kph or so, something that Hanif handily beats. Hanif also has a better handling car. But I suspect that my lap times would be quicker than his at Mosport. Why? Shear grip!

I'd love to see an analysis...

What ?!?!?!? 205 easily? In my dreams!

I hit 180ish at the top of the hump and depending on my courage that day, I'll gain another 10-12 depending on how far I want to take the braking into 8.

One slip-up in 5b and you can subtract 5kph for the entire straightaway.

I kept my foot planted and was passed as if standing still by both Christian and McAuley. Chances and the same for you if we were heading up the hill at the same time.

[Chris91GT]

What ?!?!?!? 205 easily? In my dreams!

I hit 180ish at the top of the hump and depending on my courage that day, I'll gain another 10-12 depending on how far I want to take the braking into 8.

One slip-up in 5b and you can subtract 5kph for the entire straightaway.

I kept my foot planted and was passed as if standing still by both Christian and McAuley. Chances and the same for you if we were heading up the hill at the same time.

Now, now... no bench sandbagging!

[Carguy]

Here is the classification based on IPAX averages. But hey, I'm not an IT student... :p :rolleyes: :)

[Carguy]

What ?!?!?!? 205 easily? In my dreams!

I hit 180ish at the top of the hump and depending on my courage that day, I'll gain another 10-12 depending on how far I want to take the braking into 8.

One slip-up in 5b and you can subtract 5kph for the entire straightaway.

I kept my foot planted and was passed as if standing still by both Christian and McAuley. Chances and the same for you if we were heading up the hill at the same time.

Amen to that!