NADAC Forum

General => General Discussion => Topic started by: arne on April 06, 2019, 08:12:46 PM

Title: Weavers DRI Calculation Suggested Improvement
Post by: arne on April 06, 2019, 08:12:46 PM
I really like the DRI calculation that NADAC uses. It is a tool other venues simply do not have.  Especially for the single obstacle events such as Tunnelers, Jumpers, and Hoopers I find it an excellent way to track our team's improvement over time.  As we strive to get tighter turns and more confident runs with fewer mistakes, I hope to see a gradual rise in our DRIs.

I do think the DRI calculation could be improved for Weavers. The current DRI is strongly dependent on the course length. You always have 3 sets of 12 weaves through which the dog runs at a much slower pace than the remainder of the course.  The rest of the course resembles a Tunnelers course.  If you have a short course, the slower weaves make up a larger percentage of the course so a dog's overall yps (and DRI) is going to be less than a long weavers course. The current DRI calculation does not compensate for this. Over the range of course sizes being used, I believe the current DRI calculation could be impacted by more than 10 points just because of the course length.  I think that is a big variable when you are trying to monitor improvement or performance loss.  It is a variable that does not need to be if the DRI equation for Weavers is slightly modified. The math is a little more complicated, but in the age of spreadsheets once it is set up any calculations can be done easily.  In the end most people just look at and track the final number.

There are probably multiple ways to make the improved calculation.  My suggested calculation would only require one additional piece of data -- the average run length for which the current 100 DRI yps were calculated.  Even if this is not known, you could pick a reasonable number and you would still get a substantial improvement in consistent numbers.  I would use the same calculation of DRI.  I would just adjust the 100 DRI yps to course length using the following formula:

   Adjusted DRI 100 yps= L/[(A/W) + ((L-A)/T)]
where L = Course Length (Yards)
         A = Average Weaver Course Length for Which the Current DRIs were calculated
         W = Current Weaver 100 DRI yps
         T = Current Tunneler 100 DRI yps


As an example I took the 20" dog 100 DRI yps listed in the Handbook -- 5.35 yps for Weavers and 7.35 yps for Tunnelers.  I guessed 145 yards for the average veaver course length used to calculate those 100 DRI yps.  With the current calculation you would get a 100 DRI if your dog ran at 5.35 yps, regardless of course length.  But it is going to take a much faster dog to run that on a 105 yard course than to run that on a 165 yard course.  Below shows the adjusted DRIs over the course lengths I have seen - 105 yards to 165 yards.

Course Length     Adjusted DRI
    105                     110
    125                     104
    145                     100
    165                      97   

Arne Lindberg

Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Jeff Lyons on April 06, 2019, 10:08:14 PM
Wow, an interesting concept, Arne. 

Clearly, a lot factors go into DRI for games like TnG and Weavers.  I had a slow weaving dog, and I was so grateful to see a Weavers course that had a lot of tunnel performances on it.  When hoops were introduced, it was much more difficult for her to make-up the  time she lost in the weavepoles because hoops added yardage between the obstacles, but they don't add yardage like tunnels do. In fact, there seems to be slight trend away from weavers being such a tunnel-heavy course (it is easier on the course-builders, and dogs), so I'm not sure that factoring in Tunnelers DRI yps is accurate.

However, I came to realize the real skill in the Weavers class is simply weaving.  It is a special skills class that is designed to showcase the dog's ability to weave. The dogs that weave fast and efficient can make time and even 100 DRI on shorter courses.  My hat is off to those dogs, as they do indeed have the special skill necessary to excel in Weavers. The rest of us need more obstacles that are not weavepoles to increase our DRI! I am grateful for the variety of weavepole courses some are longer, some are more tunnel-heavy.  In fairness, it is the shorter courses that really test the dog's ability and efficiency in the weaves.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: KarissaKS on April 07, 2019, 06:02:31 AM
Really, this whole concept could be applied to SCT as well. Back in the day, Luke had a hard time making time in Elite Weavers even when he was "clean" with no run-bys or pop-outs. We worked hard to improve our turns and efficiency and he eventually completed two Versatility NATCHs, but I would always groan when we saw a 13-14 obstacle Weavers course. Those are very hard to make time. But on an 18-obstacle course he would come in 5+ seconds under time. It would be nice if there was a way to balance this out a bit.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Cathie Cage on April 07, 2019, 06:39:38 AM
Sorry Jeff I have to disagree with you on this one. For years I have been having this conversation with other competitors. I have had and know of many fast dogs and when there is a short weavers course never get a DRI over 100. I have always thought there should be an adjustment made when there is a short course. The easiest way to eliminate this issue is to do away with all short weavers courses.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 06:59:39 AM
I have to think about this a bit more before we say too much.

But just some numbers to play with:

Out of 565 Weavers runs with a 100+ DRI

14% were under 125 yards
48% were between 126 and 150 yards
36% were over 150 yards

Compare that to Tunnelers where we have a little more data with 1094 runs over 100 DRI:

14% were under 125 yards
48% were between 126 and 150 yards
36% were over 150 yard

Since those numbers are exactly the same I went ahead and did Jumpers too which is:

528 runs

1.3% were under 125 yards
88% were between 126 and 150 yards
9.84% were over 150 yards.

Everyone can deduce what they want from those numbers.   


Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 07:07:42 AM
And in regards to the original question, it would just be a lot simpler to ensure that all Weavers courses fall within that 125-150 yard range.

Which is something that was going to happen anyway on January 1st once judges are designing their own courses.

Right now courses are a bit of a headache for us as we're always behind the eight ball so to speak on getting them out.  So there just isn't time to review all of these things when it's only a handful of people designing courses.    Once January 1st hits each judge will be pulling their weight a bit more in regards to course design, so we can start to look at these sort of numbers a lot more in depth and putting guidelines in place for where courses need to fall yardage wise.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Ed and Dino on April 07, 2019, 07:19:57 AM
This is interesting, I get what the original poster is getting at.
The length of 3 sets of 12 poles where collection and slower speed is needed is greater on a shorter course and therefore makes it less likely to achieve a high DRI when that number comes from an average course length where the percentage of length in weaves is less.

I disagree that the tunnelers DRI should somehow be incorporated into the calculation.
Weavers DRI should only be based on Weavers.

Chris' percentages are interesting but not really the data that is needed to see if there is an issue. if 14% of runs are shorter and 14% is 100+ DRI runs then I'd say there is no issue.

What we would really need to see is what is the overall percentage of 100+ DRI runs in a shorter course length compared to all runs at that length and then compare that to the number of 100+ DRI runs at a longer course length compared to all runs.
This would then validate if it is truly less likely to get those higher DRI runs on shorter Weavers course length.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 07:39:53 AM
This is interesting, I get what the original poster is getting at.
The length of 3 sets of 12 poles where collection and slower speed is needed is greater on a shorter course and therefore makes it less likely to achieve a high DRI when that number comes from an average course length where the percentage of length in weaves is less.

I disagree that the tunnelers DRI should somehow be incorporated into the calculation.
Weavers DRI should only be based on Weavers.

Chris' percentages are interesting but not really the data that is needed to see if there is an issue. if 14% of runs are shorter and 14% is 100+ DRI runs then I'd say there is no issue.

What we would really need to see is what is the overall percentage of 100+ DRI runs in a shorter course length compared to all runs at that length and then compare that to the number of 100+ DRI runs at a longer course length compared to all runs.
This would then validate if it is truly less likely to get those higher DRI runs on shorter Weavers course length.


Looking at all DRI's number for Weavers it looks like this:
Under 125 yards: 5.54% got over 100
Between 126 and 150 yards: 8.1% got over 100
Over 150 yards: 11.12% got over 100 DRI.

Now what is interesting there, is that DRI's were originally reserved for the top 5% of dogs in a class.

We lightened it up and made it the top 10%.

So then we're getting into a question of whether we are just continuing to lessen the standards to meet what people want.  Versus what the original goal was actually meant to be.


Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 07:46:46 AM
And for Tunnelers all dogs:

under 125:  10.23%
Between 126 and 150: 11.94%
Over 150:  9.36%

The other thing to keep in mind, there just aren't that many weavers courses that are under 125 yards.  So it's therefore going to have a smaller Q rate compared to the more common course lengths, and therefore skews things even more
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 07:49:56 AM
Weavers courses that are under 125 yards make up for 15% of the courses used.

126-150 make up for 58% of courses

Over 150 yards make up for the remaining 26%
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 08:54:11 AM
Alright last post for awhile!

I think the last thing I want to say is that sometimes people forget that agility occurs in more than their local area, and that DRI is a number based off runs around the world.

So it can be hard I know, to never get a 100 DRI and then assume it's the courses fault.   But the hard reality is that the numbers are based off dogs around the world, who may just be that incredibly fast, and the DRI numbers are reserved currently for the top 10% of dogs.    So you may have the absolute fastest dog in a specific area, but it doesn't quite make the list when compared to the other parts of the world.

And they used to be for the top 5% of dogs.   Which as Jeff mentioned, is where those shorter courses come into play since that was the original intended length for Weavers.  Not a 15 or 17 obstacle course.    We've lightened up a LOT by allowing the top 10% instead of 5%.   And we've lightened up even more by leaning towards slightly longer courses in general, as you can see in one of my above posts about the percentage of courses used and their lengths.
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Vicki Storrs on April 07, 2019, 09:34:46 AM
Okay, I CAN  have a pretty good “math hat” when necessary, but it is currently in use doing our taxes, so this is just my “first impression.”  LOL
I don’t think the original poster was trying to angle for more/easier 100 DRIs at all. When I read it I thought they were simply saying “I use the DRI numbers—a useful perk that NADAC offers—to help to evaluate my training,” whether they are in the 100s or the 60-70s (where mine usually are).  And in doing that he felt they were Less Accurate—for THAT purpose—due to the larger impact of 3 weaving opportunities (since it is often the big slow spot for so many of us) in a short weavers course as compared to a longer course. So since the course length isn’t listed on our points page he felt the Weavers DRI became less useful for his evaluation purposes. And not simply “complaining” about an issue, he also offered up a possible solution to the problem as he perceived it. IF this is an accurate interpretation of the post, and Even if it isn’t a viable solution based on worldwide NADAC statistics, KUDOS to you for your thoughtful approach!!
Vicki
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: dogrsqr on April 07, 2019, 09:43:04 AM
Do we really need DRI s to measure our training?  Can’t we keep our own records which would list the course length and then you can look back thru your weavers by length to see if you’re improving?  A simple excel spreadsheet is really all you need to measure your own progress.

Gina Pizzo
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Chris Nelson on April 07, 2019, 10:25:52 AM
I also want to say I am very happy that Arne is bringing this up.

It's a good subject and a nice one to talk about.

Whether the end result is everything stays the same, or we switch things up, it's still good to have the conversation and see what we turn up.

I had never looked up any of the percentages that I posted above until this morning, and it's some really interesting numbers that are good to look at.   

So let's keep the conversation going in a constructive way and see where it goes
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: dogrsqr on April 07, 2019, 10:57:40 AM
Sorry if anyone took my comment negatively it wasn’t meant to be.  I have kept a spreadsheet of all my dogs qualifying runs to use as information.  If you see a sudden decrease in YPS it could indicate a physical/medical issue or could help you decide when it’s time to retire or semi/retire.  It can be really tough to make those emotional decisions but when you have numbers staring you in the face it makes it tougher to ignore. It can also help you see those slight improvements that we sometimes miss because we see our dogs every day.

Gina Pizzo
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Ed and Dino on April 07, 2019, 06:14:13 PM

Looking at all DRI's number for Weavers it looks like this:
Under 125 yards: 5.54% got over 100
Between 126 and 150 yards: 8.1% got over 100
Over 150 yards: 11.12% got over 100 DRI.

Now what is interesting there, is that DRI's were originally reserved for the top 5% of dogs in a class.

We lightened it up and made it the top 10%.

So then we're getting into a question of whether we are just continuing to lessen the standards to meet what people want.  Versus what the original goal was actually meant to be.

Based on this and the post after this one where it shows for Tunnelers all lengths are closer to 10%, I would have to conclude that it is harder to get a 100+ DRI on a shorter Weavers course.

I am not advocating a change just acknowledging what Chris' great statistics show.

Thanks for sharing the data Chris, this is definitely an interesting topic that Arne brought up.

As to how to compare your times within different Weaver length courses, my math skills are not that good but if under 150 then your DRI would likely be higher on a longer course if your dog ran at it's normal speed thru course.

The DRI is an interesting number but every run / trial has variables and you just need to consider that when analyzing numbers.

Perhaps you dog is faster at inside trial, maybe faster on overcast days at outside trials. 
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: arne on April 07, 2019, 09:35:26 PM
Great discussion!  I do apologize in that I was not thinking about all the awards NADAC has tied to DRI in my original post.  I can see how any change in the calculation could ruffle some feathers.  My current dog is fast but he is not a regular member of the 100+ DRI club.  It makes no personal difference to me whether the 100 DRI is based off of the top 5% or off of the top 10%. What does make the DRI such a powerful tool is that it is based off of the fastest runs.  Chances are that the vast majority of the 100+ DRI runs did not have any off courses or bobbles or missed weave entrances.  It gives me a clear marker of what the best teams can accomplish. It is something I cannot get with only data from my dog - with lots of off courses and bobbles and missed entrances.  My goal for the post was only to improve the tool for Weavers where I thought distance had a strong influence on the current DRI calculation and really diminished its power.  I agree with Ed that the numbers Chris posted seem to support the influence of distance on the Weavers DRI.

As Chris stated, the easiest way to fix the problem would be make all of the Weavers courses the same length. My guess would be if the courses were all within a range of 25 yards, the distance influence on the Weavers DRI would probably be +/-2 or 3 from the average.  Or to state differently, the same run on the shortest allowed course might have a DRI of about 5 less that of the longest allowed course.  But that is only an estimate based on my proposed formula that uses the Tunnelers DRI 100.  As others posted, that might not be the perfect DRI 100 to use because there could also be Hoops.

I was amazed at how fast Chris came back with the various statistics from the DRI calculations.  All the data must already be in easily manipulated spreadsheet form?  If that is true, there would be no need to use a formula like I originally posted.  The rest of this post is just a proposed alternative way to calculate a Weavers DRI without a distance influence if that is the decided path.  This alternative DRI would also help get the desired percentage of dogs with a 100 DRI+ regardless of course length.  If you like the math, continue reading.  If not, don't worry about the rest of this post.

If you have all the yps data with the length of each associated run, I would make two Excel columns - one for yps and the second for the associated run length.  I would sort the data together on run length. Depending on how much data I had, I would break the data up into set run length increments (maybe 100-105, 106-110, 111-115, etc).  I would then sort each of those increments on yps.  I would take the top percentage ("top" being whatever I wanted the DRI 100 based on) and average for the DRI 100 for each increment.  I would then do a best fit with DRI 100 on the y-axis and run length on the x-axis.  (If a good best fit is a horizontal straight line, then you have proved there is no distance impact on DRI. I highly doubt that will be the case. You might need a second order fit.)  All this work is a one time (or maybe every few years) deal.  Once you have the equation of the line, you just put it in another excel speadsheet.  Then for each course you could just enter the course length into the equation and it will calculate the adjusted DRI 100 for that course.

Arne





Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: arne on April 08, 2019, 07:21:09 AM
If you are not comfortable with finding the equation of a best fit curve, you could get similar results with a long IF statement plugging in the calculated DRI 100 for each increment of course length. 
Title: Re: Weavers DRI Calculation Suggested Improvement
Post by: Audri, Cee Cee, Lily, Toto, and Calypso on April 08, 2019, 09:08:43 AM
I have to think about this a bit more before we say too much.

But just some numbers to play with:

Out of 565 Weavers runs with a 100+ DRI

14% were under 125 yards
48% were between 126 and 150 yards
36% were over 150 yards

Compare that to Tunnelers where we have a little more data with 1094 runs over 100 DRI:

14% were under 125 yards
48% were between 126 and 150 yards
36% were over 150 yard

Since those numbers are exactly the same I went ahead and did Jumpers too which is:

528 runs

1.3% were under 125 yards
88% were between 126 and 150 yards
9.84% were over 150 yards.

Everyone can deduce what they want from those numbers.   

While I think that the numbers are interesting, I think what Arne was trying to say is that in weavers, a shorter weavers course is made up of a lot more ground on weaves, which for any dog, is slower than simply running through a tunnel.  I get where he is coming from as a tool to look at my progress.  I use them for the same reason.  Whoever wrote that we could all do our own, yeah that is awesome if I had all the time in the world.  I don't and I run 4 dogs, so the amount of data entry after each trial would be a lot since I generally run 40 events in a typical weekend.  The DRI is a nice, easy way to compare from time to time.  In fact, I just did a graph (because I am a numbers geek and because the numbers were readily available) on one of my dogs to compare all of her DRI's over the past 8 years to see how she is holding up as she turned 12 today.  Her DRI's have actually stayed pretty consistent on the events I did, so it was interesting to see.   Maybe instead of adjusting the DRI, YPS could be on the download?