You are a techno moron. You think you know things but you don't understand the underlining data and numerological methods. And you end up looking like an idiot with your pants down around your ankles. You are in way over your head. You know nothing. Give up you dufus.
SW and AquaDyne. As I just posted on the other thread (in order to get AquaDyne's attention, to input on this thread if necessary), the fact is that Mimi's Cadence Data is being subjected to a 'rounding artifact', as explained by AquaDyne on the other thread. Since this 'artifact' exists in the raw data as uploaded to Strava, it suggests then that a data operation introducing the 'rounding artifact' was performed on the data prior to it being uploaded to Strava. SW has so far determined other Suunto users do not have such artifact, which potentially suggests external 'data tampering' of the Mimi data prior to it's being uploaded.
“When the debate is lost, slander becomes the tool of the loser.”
? Socrates
Sandy? wrote:
I found this random Suunto runner on Strava, and his cadence data shows a somewhat similar pattern of missing cadence values as Mimi's data:
https://www.strava.com/activities/1187449540
Good find, it does look like this George with a Suunto Spartan Sport also shows the same type of Swiss cheese cadence data. Could it possibly be a problem specific to Suunto Spartan watches and not older Suunto models? I think we need to see if we can find data from other Suunto Spartan Ultra or Suunto Spartan Sport users since this George only one one run in his Strava from Sept. 14th.
I'm uploading analyzed files for Suunto Spartan Ultra or Suunto Spartan Sport data that I find to this sub-folder on my Google Drive:
https://drive.google.com/drive/folders/0B1F3plcU7fEmWThENEhwYk1xeXcI overlooked George's older runs. He has older runs on his Strava. I will also analyze.
Suunto George's cadence data isn't quite analogous to Mimi's, at least for the specific run I checked, since there are only a very small number of cadence values outside his typical running cadence. But for those outlying values, it does show a similar swiss cheese hole pattern. Here's one of George's histograms, with the vertical scale zoomed way in:
I've found this run from a Jeff using a Suunto Spartan Ultra that doesn't show the Swiss cheese behavior, maybe his run is too short and he runs too fast:
I've analyzed the data from a Ted that uses a Suunto Spartan Ultra on Strava and it also shows the same Swiss cheese cadence behavior. Note that Ted's Strava has the cadence in RPM instead of SPM, hence why it is so high:
the first "something to consider" wrote:
SW and AquaDyne. As I just posted on the other thread (in order to get AquaDyne's attention, to input on this thread if necessary), the fact is that Mimi's Cadence Data is being subjected to a 'rounding artifact', as explained by AquaDyne on the other thread. Since this 'artifact' exists in the raw data as uploaded to Strava, it suggests then that a data operation introducing the 'rounding artifact' was performed on the data prior to it being uploaded to Strava. SW has so far determined other Suunto users do not have such artifact, which potentially suggests external 'data tampering' of the Mimi data prior to it's being uploaded.
I've uploaded all the Suunto Sparta Ultra/Sport watch data from random people onto my Google Drive along with their cadence histograms. It looks like the issue with Mimi's cadence data is indeed because of a extremely odd problem specific to all Suunto Spartan watches and not because of any doctoring. I don't see why Suunto would do this intentionally to one of their flagship products, it seems like a major bug.
I think Mimi's Strava data has passed my scrutiny and it is fair to say now that the Strava data looks authentic.
https://drive.google.com/drive/folders/0B1F3plcU7fEmWThENEhwYk1xeXcscam_watcheroo wrote:
I think Mimi's Strava data has passed my scrutiny and it is fair to say now that the Strava data looks authentic.
Note that this just leaves the matter of verifying the uncut video of Mimi running to check that the Strava data matches what is seen in real life with the same cadence and same durations.
scam_watcheroo wrote:
I think Mimi's Strava data has passed my scrutiny and it is fair to say now that the Strava data looks authentic.
Is there anything in the cadence data that shows if the knee/back injuries caused any changes. If yes, then that could be very useful for runners and coaches to catch problems before they get worse.
scam_watcheroo wrote:
It looks like the issue with Mimi's cadence data is indeed because of a extremely odd problem specific to all Suunto Spartan watches and not because of any doctoring. I don't see why Suunto would do this intentionally to one of their flagship products, it seems like a major bug.
Yeah. I was thinking more about this. We're looking at data with with 1 second resolution, with a separate cadence measurement for each data sample. How does that even work? The sampling rate is nearly as fast as the cadence events being measured. There's only going to be 1 or 2 arm swings in a 1 second period, so there's no way to get an accurate cadence measurement just by counting arm swings. I suppose you could measure the elapsed time of a single arm swing, but I think that would be highly variable.
I'd argue it almost doesn't make sense to think of cadence on a second-by-second basis, unless you're talking about a moving average from that instant going back ~15 seconds in time. In order for the measurements to be meaningful, cadence really needs to be averaged over a time interval containing at least a couple of dozen steps. So I'm not too surprised if Suunto Spartan cadence data looks wacky at 1 second resolution. It would be nice if they fixed it though, since other GPS watches don't seem to have the same problem.
scam_watcheroo wrote:
I think Mimi's Strava data has passed my scrutiny and it is fair to say now that the Strava data looks authentic.
Yeah. It'd take quite the hacker to figure out a cadence artifact particular to one model of the watch and reproduce that on spoofed data.
Glad we've gotten to the bottom of this and removed doubt about this data.
scam_watcheroo wrote:
Note that this just leaves the matter of verifying the uncut video of Mimi running to check that the Strava data matches what is seen in real life with the same cadence and same durations.
Huh? First you say it's authentic and now you insist on more proof? When will you stop raising the bar?
She's not trying to claim a record. She has nothing to prove to anyone. So far everything we've seen says her run, such as it was, was legit. Let's let her recover and wish her the best.
We should not just compare models but also firmware versions, something which public data aggregation sites like Strava will not reveal.
In "embedded electronics" world cadence works with a gyroscope sensor, the same stuff that is in a cellphone for games or "level meters".
As someone who works with ATM software across all layers I would think that scam-watcheroo should just open a bug report with Suunto, or someone with such a product (if a product registration or such is needed).
This needs to go through the manufacturer, all this reverse engineering is nice hobbyist (read enthusiasts) level but meybe energy is better spent to find the Product Manager for these watches and have him relay with their development.
Suunto 4 Injuries wrote:
Is there anything in the cadence data that shows if the knee/back injuries caused any changes. If yes, then that could be very useful for runners and coaches to catch problems before they get worse.
You could always do a ratio of the time spent at a fast run cadence over the slow run/walk cadence. But this could vary due to weather conditions and how hilly the terrain is.
Looking through Mimi's cadence histograms later in her runs, you can see obvious days where she spent pretty much most of the time at the walking cadence with very low high cadence counts. And this isn't in the mountains or extreme weather that Sandra had to face where we see legitimate need for a drop in cadence.
Look at the cadence histograms for her 10-02, 10-03, 10-05, 10-07, and 10-09 runs where she unexpectedly spent the afternoons just at waking cadence. That is a pretty obvious sign something is wrong, especially since we know that this is relatively flat terrain. Mimi then goes back to her fast cadence running for the remainder of October as seen in the histograms, which is what finally did her knee in.
This was also preceded by hospital visits as well as the obvious video signs I pointed out way back of Mimi having trouble even walking up our down a few RV stairs. There was also Mimi not being able to stand straight with her knees locked straight. There were plenty of signs something was not right and LRC pointed them out but it looks like URC and Mimi and her team/supporters just tried to ignore/hide the obvious signs.
AquaDyne wrote:
Huh? First you say it's authentic and now you insist on more proof? When will you stop raising the bar?
She's not trying to claim a record. She has nothing to prove to anyone. So far everything we've seen says her run, such as it was, was legit. Let's let her recover and wish her the best.
I'm not raising the bar, I have been insisting for the release of uncut video from Mimi and her team pretty much from the start of all the suspicion. Check the original thread. It is because they didn't release uncut video that all this other analysis is necessary.
Thus was one of the first Google results for "Arduino accelerometer" and gives a good idea how this stuff works:
https://forum.arduino.cc/index.php?topic=58048.0
Here another forum post claiming Suunto issues with accelerometer-based data recording:
http://forums.watchuseek.com/#/topics/918495
(I am not saying this proofs/disproves anything, I am just the same data (and data logger) geek as many in here.)
Strange stuff?
Sandy? wrote:
I'd argue it almost doesn't make sense to think of cadence on a second-by-second basis, unless you're talking about a moving average from that instant going back ~15 seconds in time. In order for the measurements to be meaningful, cadence really needs to be averaged over a time interval containing at least a couple of dozen steps. So I'm not too surprised if Suunto Spartan cadence data looks wacky at 1 second resolution. It would be nice if they fixed it though, since other GPS watches don't seem to have the same problem.
Most likely watches maintain some type of moving window for the cadence data to hold a certain amount of time off accelerometer data from which to find the minima/maximas in the sinusoidal like curves. Suunto should really fix what looks to be like a logic error in their Spartan watches since it does happen in older models.
European____ wrote:
Here another forum post claiming Suunto issues with accelerometer-based data recording:
That thread is deal with cumulative integration error from going from acceleration data to distance. It is different from the issue we have here which doesn't involve integration.