Unless you are at a level where IAAF starts to keep track of your PBs, it is up to you to establish your criteria for PRs.
Unless you are at a level where IAAF starts to keep track of your PBs, it is up to you to establish your criteria for PRs.
I've got a better one. No gps data, just a selfie and a recap of the time. My overweight friend is trying to improve his mile and a half time for the military posted a selfie that says "5.5 mile run done in 30 minutes. Caught my second wind after the 4th mile and sprinted the last mile in the 4 minute range"...... ugh..... I'm going to need to see some gps data on that.......
Interesting Stuff wrote:
I don't think pop_pop or GIS really understand this GPS thing after all.
Dude, you don't understand it. If my GPS says a I ran a 10K on a training run in 35:00, I know the time, but I don't know the real distance.
For examples, 35:00 is 2100 seconds, I could have ran anything from 0K to 31.5K.
Here are some real-world GPS numbers on an election day when I'm waiting and hoping that the heavy rain will stop before I head out. I run past my polling place and am trying to decide whether to stop in, dripping wet, and vote. I think not.
For a couple of years I've been using the same "8.55 mile" course a few times a week -- starting and stopping at the same places, very rarely crossing the road or drifting out. It's a winding loop (well, a lollipop course I suppose) with an elevation differential of about 860 feet.
First with a Garmin 910, then with a 920.
N = 57
Range: 8.50 - 8.61
Average = 8.55 miles
Standard deviation = .0242 miles, or 128 feet
Standard error = about 17 feet
If I toss out the two highest and two lowest observations, the STD falls to about 110'.
Is this a properly-controlled experiment? Of course it isn't. But it strongly implies that when I go out on that course, there's a 99% chance that my true distance (plus/minus 3 standard deviations) lies between 8.48 and 8.62 miles, thus about .15 mile of uncertainly in there. Per mile, that's roughly .014 mile, roughly 90 feet.
I'd argue that my GPS gives me 90' per mile uncertainty, max.
In the real world, do I care? Not at all. All I care about is the elapsed time for my course.
As for the PR business -- I'd be OK with a non-race GPS PR so long as the guy doesn't start talking about how he beat his PR by 10 seconds, or anything within a reasonable range of distance error. But if it's a minute or two -- I'd give it to him. Why not?
Yeah so you didn't read my post. Here I'll make it easy for you: Smoothing exists.
A runner & PT I know sent me his Garmin reading of 31:40. His best in a race 35xx
I run with a Garmin Fenix. It automatically captures my "PRs" at various distances (5K, 10K, half and full marathon) and displays them on my Garmin Connect page. That is based entirely on the distance and time recorded by my Fenix, regardless of where and when I run it. My Fenix doesn't know whether or not I am in a race. It doesn't seem out of the ordinary that other GPS products would do similar things.
So, if your friend has his GPS unit set to automatically post his runs to social media, then the GPS may have posted the run as a "PR" without any input from him.
I don't think it should count, because it could possibly include rest. If i did a 4 x mile workout at 5min pace and paused the watch during the rest my garmin will say i ran 4 miles in 20 mins, though the elapsed time could be close to 30 mins.
Interesting Stuff wrote:
I don't think pop_pop or GIS really understand this GPS thing after all....
.
Haha. Yeah, I've never coded GIS apps before. Ever.
Help me with this smoothing. I need details. Is the device smoothing? strava? Who is smoothing? What method?
Interesting Stuff wrote:
Strava also has this other amazing thing where they take the "popular" track data and use that to fine-tune yours to... get this.... eliminate bad data.
Now you are making stuff up. If they can "smooth" as you claim, then why are there dozens of unique segments of the same cycle/running path with the same start and end points?
pop_pop!_v2.2.1 wrote:
Interesting Stuff wrote:I don't think pop_pop or GIS really understand this GPS thing after all....
.
Haha. Yeah, I've never coded GIS apps before. Ever.
Help me with this smoothing. I need details. Is the device smoothing? strava? Who is smoothing? What method?
Interesting Stuff wrote:
Strava also has this other amazing thing where they take the "popular" track data and use that to fine-tune yours to... get this.... eliminate bad data.
Now you are making stuff up. If they can "smooth" as you claim, then why are there dozens of unique segments of the same cycle/running path with the same start and end points?
He might be talking about a Kalman filter (but he's probably taking a wild guess due to the fact he didn't clarify). But even then it's ridiculous to say that this is without error, and even more ridiculous to claim that GPS is just as accurate as using a wheel.
Races use bike computers and a measuring wheel for this reason. We haven't even touched on the fact that GPS is just a simple XML file and can easily be manipulated. It's no substitute for a race, and like you I'm also an app developer.
GPS is a great guide, but not the bible of accuracy a lot of runners think it is.
I get not wanting to pay money to run a certified 5k or 10k if you're just training for yourself and not at a really competitive or noteworthy level, and so I can see why someone would want to declare a PR without having chip timing or official results to back it up, but at least run the thing on a track and ignore GPS distance.
GIS wrote:
wut wut wrote:15 meters per second? Four minute mile pace is 6.67 meters per second. What kind of math is this?
Last two 5Ks I ran on certified courses, GPS showed 3.09 and 3.12.
On a very simple level the kind of math is a combination of algebra and geometry (above most of LRC's heads). The earth is not flat*. GPS estimates are the result of triangulation* so at any given moment if the transmission is off you might be allocated 15m or more from your actual position.
Where did I get the 15 m/s margin of error? Class. I went to a top 25 institution and in my GIS courses this figure was beat into our heads. It was repeatedly asked on homework, quizzes and exams to the point where it became a gimme. I think there was an overall point conveyed by its frequent appearances in graded material.
Should get your money back from that instituation, 15 m/s is the accuracy of your spot on the globe (which is now 4-8m) that error changes very little from spot to spot (at least over the time range of a run) for your watch all that matters is your relative change from spot to spot which is pretty darn accurate.
The biggest problem for gps watches is measuring the curves and turn around. If you run a straight line your watch is going to be pretty darn accuarate. If your running back and forth or turning a lot then not so much.
Bubbadeeboos wrote:
Going from memory, the 15m accuracy is the theoretical accuracy of the pseudorange solutions used by cheap GPS receivers. In practice, many of the sources of error are correlated so something near 8m is expected with clear skies and good satellite coverage. A couple things about watches make things worse (crappy antennas, your body blocking half the sky --> fewer satellites and worse geometry...) , I imagine most watches/apps apply a smoothing algorithm, & possibly use accelerometers to minimize problems from dropped signals etc.
I use an iPhone with strava. The distance measurements don't seem too bad, but the instantaneous pace is atrocious.
I did a 5k, official time of ~18:50 but strava noted an 18:05 5k... Not sure if strava just thinks a 5k is three miles or if the gps was that wrong. A similar thing happens when I run on a track that has a strava segment on it... it extrapolates my lap times to ridiculous mile paces.
GPS wrote:You are saying a GPS is inaccurate by 'at least 15 meters'? Over the course of a 10K? That's plenty accurate to call a time/distance run legit. For a 35 minute 10k runner that's a matter of a few seconds.
I've had issues with STRAVA too. I did the Columbus Marathon in an official time of 1:29 but STRAVA said 'new marathon PR in 1:26:xx' - that's a pretty big difference. I think the Columbus course was long by a little bit, since it's official, but still...
GIS wrote:
They just keep coming. The idiocy of an app-for-everything world.
You CANNOT use commercial GPS to accurately measure a running distance. It does not work. It is in accurate by at least 15 m/s. Trees obsfucate GPS transmission. Commerical GPS can only get down to 15ft above the ground if nothing else comes first.
Jones counter. Steel tape. Chains. Even a Keson wheel. But not GPS.
GPS does not have a lot of precision.
But don't GPS watches usually make mistakes in giving a readout that is shorter than the actual distance covered?
It is pretty normal if you voted!
Nope. The Joke is on the rest of you trolls grasping at straws and hoping to get something to stick. Once again, you cannot rely on commercial GPS when precision is important. Go measure your local 400m track to experiment. You'll see what I mean.P.S. I would trust employees from a state cartographer's office and educators with their PhD in GIS over LRC troll tales any day.Back to your mom's basement!
The joke is on us wrote:
GIS wrote:On a very simple level the kind of math is a combination of algebra and geometry (above most of LRC's heads). The earth is not flat*. GPS estimates are the result of triangulation* so at any given moment if the transmission is off you might be allocated 15m or more from your actual position.
Where did I get the 15 m/s margin of error? Class. I went to a top 25 institution and in my GIS courses this figure was beat into our heads. It was repeatedly asked on homework, quizzes and exams to the point where it became a gimme. I think there was an overall point conveyed by its frequent appearances in graded material.
Should get your money back from that instituation, 15 m/s is the accuracy of your spot on the globe (which is now 4-8m) that error changes very little from spot to spot (at least over the time range of a run) for your watch all that matters is your relative change from spot to spot which is pretty darn accurate.
The biggest problem for gps watches is measuring the curves and turn around. If you run a straight line your watch is going to be pretty darn accuarate. If your running back and forth or turning a lot then not so much.
The track is the ABSOLUTE WORST place to test a gps. Any idiot knows that turns are where the most error would be and you are turning more than half the time on a track.
While the gps is not exactly accurate, you don't just add up the error from every update. It would be interesting to see a gps readout from a trail race I ran last weekend on mountain bike trails that probably had 50-100 turns per mile for the first 6 miles. I've had 3 Garmins over the years but I rarely use one anymore. I took it out last week on a trail run and it shut off about 2/3rds through the run. Probably couldn't pick up a signal for too long. I didn't realize it until I looked at it a couple miles later. Luckily I also wore a regular watch so I had my running time for the whole run.
high hanging fruit..... wrote:
The track is the ABSOLUTE WORST place to test a gps. Any idiot knows that turns are where the most error would be and you are turning more than half the time on a track.
While the gps is not exactly accurate, you don't just add up the error from every update. It would be interesting to see a gps readout from a trail race I ran last weekend on mountain bike trails that probably had 50-100 turns per mile for the first 6 miles. I've had 3 Garmins over the years but I rarely use one anymore. I took it out last week on a trail run and it shut off about 2/3rds through the run. Probably couldn't pick up a signal for too long. I didn't realize it until I looked at it a couple miles later. Luckily I also wore a regular watch so I had my running time for the whole run.
I don't think the guy who suggested the track was talking about using your GPS there. The way I interpreted it was that it would be the most accurate way to run a PR if you just took your watch there and timed yourself (no GPS!). Because you already know the exact distance of each lap.
So it seems like GPS is even more inaccurate than what I originally thought, and the only way to CLAIM a PR is in a race, and possibly on training run on the track.