2005 Season Stats...

Emmie Fisher · **Posted:** Thu Dec 15, 2005 8:58 pm

Richard Casto wrote:

I am putting all of the data into a database so I can generate some stats. Should be done tomorrow afternoon sometime. Stay tuned.

PS: Once in the DB I should be able to provide files (CSV, Excel, etc.) of the raw data to anyone that wants to do their own stats.

Are you using the "Get External Data" feature of Excel? Will need a little massaging to separate the penalties from the times. Otherwise, quick look appears to grab the data pretty well.

Richard Casto · **Posted:** Thu Dec 15, 2005 9:47 pm

Emmie Fisher wrote:

Richard Casto wrote:

I am putting all of the data into a database so I can generate some stats. Should be done tomorrow afternoon sometime. Stay tuned.

PS: Once in the DB I should be able to provide files (CSV, Excel, etc.) of the raw data to anyone that wants to do their own stats.

Are you using the "Get External Data" feature of Excel? Will need a little massaging to separate the penalties from the times. Otherwise, quick look appears to grab the data pretty well.

I was basically copy-n-pasting the HTML table into Excel and then converting (via manual clean up) that into a basic spreadsheet. I had started to write some Excel functions to break out the cone penalties, DNF and DNS stuff. But I am not an Excel expert and became frustrated with that so I just pumped the raw data into MS SQL Server. I then have some T-SQL scripts that parses apart the per run time and penalty info as well as stores each run in it's own row. I now have raw and adjusted times as well as individual cone penalties and flags for DNF and DNS. My only assumption right now is that the time are listed as "raw+penalty". So a "50.2+1" is "52.2 adjusted". I need to go back and look over data to make sure this is right.

At this point it is pretty easy to generate some interesting numbers. I have three events in so far. I just did three to make sure I could do it and now I am going to do the rest of the events.

Richard Casto · **Posted:** Mon Dec 19, 2005 3:59 pm

Ok, I finally have my data in a DB and have been able to crunch some numbers. It took so long as I ended up redesigning my DB and import process about half the way through.

Top PAX and fast on their first run

This was done by limiting the results to those whos average PAX percentage from their best five events was better than 30%. Or in other words, if you are in this list you PAXed on average in the top 30% in your best five events. I then calculated the percent difference between the first and best run. With a qualification that your "best" run couldn't be your first (otherwise you would appear to be VERY fast out of the gate, but then to never make any improvements). A number very close to 100% means that your first run was almost as quick as your best. This was done with Adjusted (cones count) times and not Raw or PAX times.

First value is % diff between 1st and best run
Second value is the PAX %

Code:

64   7.15    Kevin Allen
32   17.18   Miles Beam
73   21.71   Rick Butters
70   15.55   Patrice Bousquet
68   14.69   Rob Harvey
66   1.96    James Feinberg
24   19.19   Kevin Butler
20   5.59    Dustin Fredrickson
98   4.34    Mike Whitney
62   14.95   Donna Frank
25   9.60    Dan Ecclestone
24   10.10   Marcus McRae
23   3.56    Brian Herring
18   16.38   Bill Fitzgerald
04   29.53   Richard Casto
99   29.19   Matthew Roush
78   3.62    Scott Johnson
66   5.19    Mary E. Fisher
37   22.18   Tonya Mauldin
36   7.30    Keith Quistorff
89   2.44    Carl Fisher
76   19.74   Vincent Keene
41   7.15    Jason Mauldin
12   10.01   Sally Johnson
11   23.96   Rob Lupella
89   16.67   Dick Rasmussen
76   23.71   Arthur McDonald
53   27.23   Stephen Westerfield
51   27.05   Ryan Holton
48   15.42   Les Davis
95   23.26   Pj Aspesi
94   16.04   Charles Frank
09   15.90   Jason Panciera
65   8.08    Jim Pastorius
37   29.31   Randy Melton

Same thing, but only with top 10% PAX

Code:

64   7.15    Kevin Allen
66   1.96    James Feinberg
20   5.59    Dustin Fredrickson
98   4.34    Mike Whitney
25   9.60    Dan Ecclestone
23   3.56    Brian Herring
78   3.62    Scott Johnson
66   5.19    Mary E. Fisher
36   7.30    Keith Quistorff
89   2.44    Carl Fisher
41   7.15    Jason Mauldin
65   8.08    Jim Pastorius

I haven't tried to double check this stuff so don't bitch if it is wrong :lol:

If anyone has any other weird requests let me know.

Carl Fisher · **Posted:** Mon Dec 19, 2005 5:14 pm

Hmm- this doesn't distinguish people who are "fast out of the gate" from people who are just consistent, regardless of speed. I think you would need to consider this "first-to-fast" percentile value together with the PAX ranking (more like its inverse) to come up with an overall factor. So someone who was inconsistent but fast might score lower than a very consistent, slightly slower, driver.

But I think you should be calculating based on raw times, as that "fast out of the gate" performance means taking chances, and therefore a high risk of cone penalties. These drivers are counting on the fact that only the best run is scored- that's integral to their plan.

Brian Herring · **Posted:** Mon Dec 19, 2005 5:55 pm

Richard Casto wrote:

Ok, I finally have my data in a DB and have been able to crunch some numbers. It took so long as I ended up redesigning my DB and import process about half the way through.

Top PAX and fast on their first run

This was done by limiting the results to those whos average PAX percentage from their best five events was better than 30%. Or in other words, if you are in this list you PAXed on average in the top 30% in your best five events. I then calculated the percent difference between the first and best run. With a qualification that your "best" run couldn't be your first (otherwise you would appear to be VERY fast out of the gate, but then to never make any improvements). A number very close to 100% means that your first run was almost as quick as your best. This was done with Adjusted (cones count) times and not Raw or PAX times.

First value is % diff between 1st and best run
Second value is the PAX %

Code:

64   7.15    Kevin Allen
32   17.18   Miles Beam
73   21.71   Rick Butters
70   15.55   Patrice Bousquet
68   14.69   Rob Harvey
66   1.96    James Feinberg
24   19.19   Kevin Butler
20   5.59    Dustin Fredrickson
98   4.34    Mike Whitney
62   14.95   Donna Frank
25   9.60    Dan Ecclestone
24   10.10   Marcus McRae
23   3.56    Brian Herring
18   16.38   Bill Fitzgerald
04   29.53   Richard Casto
99   29.19   Matthew Roush
78   3.62    Scott Johnson
66   5.19    Mary E. Fisher
37   22.18   Tonya Mauldin
36   7.30    Keith Quistorff
89   2.44    Carl Fisher
76   19.74   Vincent Keene
41   7.15    Jason Mauldin
12   10.01   Sally Johnson
11   23.96   Rob Lupella
89   16.67   Dick Rasmussen
76   23.71   Arthur McDonald
53   27.23   Stephen Westerfield
51   27.05   Ryan Holton
48   15.42   Les Davis
95   23.26   Pj Aspesi
94   16.04   Charles Frank
09   15.90   Jason Panciera
65   8.08    Jim Pastorius
37   29.31   Randy Melton

Same thing, but only with top 10% PAX

Code:

64   7.15    Kevin Allen
66   1.96    James Feinberg
20   5.59    Dustin Fredrickson
98   4.34    Mike Whitney
25   9.60    Dan Ecclestone
23   3.56    Brian Herring
78   3.62    Scott Johnson
66   5.19    Mary E. Fisher
36   7.30    Keith Quistorff
89   2.44    Carl Fisher
41   7.15    Jason Mauldin
65   8.08    Jim Pastorius

I haven't tried to double check this stuff so don't bitch if it is wrong :lol:

If anyone has any other weird requests let me know.

Woah... so does all this mean that THSCC gave me an "A" on my driving? I think this may be the first time I have had a 4.0 in a LONG time

- dow
Ps. What does that second row 'really mean'. I am at work and correlation usage of my brain to numbers is quite low at the moment. Thanks!

PPs. Nevermind.. I got it. Too bad that on quite a few events, by first runs WERE my best runs. Oh well.

Kevin Allen · **Joined:** Thu Aug 28, 2003 6:42 pm **Posts:** 3479

Quote:

Hmm- this doesn't distinguish people who are "fast out of the gate" from people who are just consistent, regardless of speed.

Consistency = good for rallyx. All you people at the top of that list need to come play in the dirt.

Kevin Allen · **Joined:** Thu Aug 28, 2003 6:42 pm **Posts:** 3479

Quote:

So someone who was inconsistent but fast might score lower than a very consistent, slightly slower, driver.

Or maybe the "slower" driver was really in a slow car, but drove it right at the limits consistently.

Richard Casto · **Posted:** Mon Dec 19, 2005 6:13 pm

Carl Fisher wrote:

Hmm- this doesn't distinguish people who are "fast out of the gate" from people who are just consistent, regardless of speed. I think you would need to consider this "first-to-fast" percentile value together with the PAX ranking (more like its inverse) to come up with an overall factor. So someone who was inconsistent but fast might score lower than a very consistent, slightly slower, driver..

Carl, I understand the comment about people who are just consistent, but not sure about the comment "regardless of speed". That is why only those who consistently PAXed high are in the list. I did look at the entire list and you will find people who get closer to the magical 100%, but you also see people who are very far down on the PAX list. As well as people who have later runs that are slower than their first (but are still close to 100%, they are just 101%, etc.) Some people ARE more consistent than the top PAX folks, but the are also driving slower (which is what I think you are talking about).

I picked the 30% cut off because frankly that is where I made the cut off.

The 10% PAX obviously IMHO shows some top drivers.

Now trying to combine both this first/fast metric and PAX into a single one is not something I am going to attempt.

If "PAX sucks" then PAX + something else is also going to suck.

Carl Fisher wrote:

But I think you should be calculating based on raw times, as that "fast out of the gate" performance means taking chances, and therefore a high risk of cone penalties. These drivers are counting on the fact that only the best run is scored- that's integral to their plan.

I will do one based upon RAW times. One problem IMHO with RAW times is that if someone ends up significantly shortcutting an element (not so bad it is a DNF) it can greatly inflate that time. But then again, if they get that coned but fast run early AND they are able to clean it up enough that they can still beat that initial raw time, then I think that is what you are looking for. I will put that together tomorrow.

Richard Casto · **Posted:** Mon Dec 19, 2005 6:30 pm

Ok, same thing as before, but with RAW times. So if you were being risky in your initial runs and then cleaning it up later, you will move up the list. If you were being risky and really blew it you will move down the list. OR this could just be a load of crap statistical mumbo jumbo. You decide...

Top 30% PAXers over best 5 events

Code:

57   27.23   Stephen Westerfield
48   7.15    Kevin Allen
02   5.19    Mary E. Fisher
89   3.56    Brian Herring
77   7.30    Keith Quistorff
62   29.53   Richard Casto
60   14.95   Donna Frank
53   4.34    Mike Whitney
53   9.60    Dan Ecclestone
32   22.18   Tonya Mauldin
26   5.59    Dustin Fredrickson
24   10.10   Marcus McRae
23   14.69   Rob Harvey
22   29.19   Matthew Roush
22   1.96    James Feinberg
14   21.71   Rick Butters
95   17.18   Miles Beam
93   8.08    Jim Pastorius
87   19.19   Kevin Butler
75   2.44    Carl Fisher
53   7.15    Jason Mauldin
47   16.04   Charles Frank
28   15.42   Les Davis
28   23.96   Rob Lupella
24   16.38   Bill Fitzgerald
18   15.55   Patrice Bousquet
72   19.74   Vincent Keene
63   3.62    Scott Johnson
48   27.05   Ryan Holton
82   15.90   Jason Panciera
26   10.01   Sally Johnson
72   16.67   Dick Rasmussen
67   23.26   Pj Aspesi
57   23.71   Arthur McDonald
56   29.31   Randy Melton

Woo Hoo! I am up to 6th. I will keep tweaking the numbers until I can prove I am the best driver on the planet. :roll:

Top 10% PAXers over best 5 events

Code:

48   7.15    Kevin Allen
02   5.19    Mary E. Fisher
89   3.56    Brian Herring
77   7.30    Keith Quistorff
53   4.34    Mike Whitney
53   9.60    Dan Ecclestone
26   5.59    Dustin Fredrickson
22   1.96    James Feinberg
93   8.08    Jim Pastorius
75   2.44    Carl Fisher
53   7.15    Jason Mauldin
63   3.62    Scott Johnson

robharvey · **Posted:** Mon Dec 19, 2005 9:10 pm

Richard Casto wrote:

The 10% PAX obviously IMHO shows some top drivers.

Or drivers that have a kick-ass PAX!! :angeldevil:

Richard Casto · **Posted:** Mon Dec 19, 2005 11:09 pm

robharvey wrote:

Richard Casto wrote:

The 10% PAX obviously IMHO shows some top drivers.

Or drivers that have a kick-ass PAX!! :angeldevil:

PAX sucks right!

Seriously you have to take all of this with a grain of salt as the only thing that maters at the end of the day is the class results for each event. But this is the off season, so we have to do something to keep ourselves occupied.

Anyhow, I can easily recalculate all of last years results with the 2006 PAX!! Who thinks it's going to do much? Considering that pretty much everything (that runs in our club) other than STS and STS2 has a harder PAX, I don't think it will do anything other than make me and Jim F. look better! Jim doesn't need a softer PAX. :shock:

MikeWhitney · **Joined:** Wed Aug 20, 2003 9:15 pm **Posts:** 4724

Hey Richard - if you're taking requests

How about standard deviation of raw times, ignore cone counts, ignore DNF runs, averaged over all events...

Like Brian, I have had a few events with the first run fastest. And like you I want to keep tweaking until I'm proven the best

Brian Herring · **Posted:** Tue Dec 20, 2005 12:17 am

MikeWhitney wrote:

Hey Richard - if you're taking requests

How about standard deviation of raw times, ignore cone counts, ignore DNF runs, averaged over all events...

Like Brian, I have had a few events with the first run fastest. And like you I want to keep tweaking until I'm proven the best

I dont think there are any arguements about who is the fastest drivers in the club are... they prove it over the long term.

- dow

Kevin Butler · **Posted:** Tue Dec 20, 2005 1:30 am

Richard,

Too much free time on your hands!

Good stuff though.

The way I interpret the results for myself is

A. I'm consistent within less than 4% wether raw or pax
B. I don't learn enough, fast enough, hence the consistency within 4%
C. I'm predictable as hell, like Jim P told me all year.

I would consider myself to be one of the "consistent, slightly slower drivers" referred to by Carl Fisher (no offense taken). The surprising thing was that I creeped into the upper 20th percentile of pax standings.

Using the same statistics, my goals for 2006 would be to

A. Increase the % difference between my first and fastest runs. To me,
that would be an indicator of better learning.
B. Improve my overall pax standing from 2005's

David Teague · **Posted:** Tue Dec 20, 2005 8:49 am

robharvey wrote:

Richard Casto wrote:

The 10% PAX obviously IMHO shows some top drivers.

Or drivers that have a kick-ass PAX!! :angeldevil:

It is aslo course dependent, at sanford in may Mary, Carl and I were all in the top 15 in pax with softest pax (HS). However at larunburg, we all fell down in the mid 20's and 30's. I belived this happen because of the several very slow elements at the Larenburg event compared to a more flowing course at Sanford.

THSCC Forums

2005 Season Stats...

Who is online