Incorrect xpower values
Mattyb
I have a ride which just seems to give me the wrong data.
Powertap said the average power is 259w for 35mins or so
WKO and the new poweragent agree with the average and give normalized
power of 270w or so
GC reports an average of 252w and xpower of 246w
My peak 230min power is reported as 268w so something is wrong here.
I have noticed that the interval is showing total time, not ride time.
This route had quite a few traffic light stops and was very hilly so
it's not a isopower effort.
Something isn't right and I suspect that time at 0 speed is being
factored into the numbers.
Doesn't make a difference how I bring the file in - via PT download,
CSV, WKO etc. numbers = the same
Sean Rhea
Can you send the file in question to the list? Also, you're talking
about the numbers reported on the Ride Summary tab, right? Not those
in the Metrics tab?
Sean
> --
> _______________________________________________
> Golden-Cheetah-Users mailing list
> golden-che...@googlegroups.com
> http://groups.google.com/group/golden-cheetah-users?hl=en
--
"I refuse to accept the idea that the ‘isness’ of man’s present nature
makes him morally incapable of reaching up for the eternal ‘oughtness’
that forever confronts him." --MLK
Mattyb
I've also noticed that todays ride does a similar thing where the
interval shows the total time rather than the ride time without 0
speed data.
I'll upload the files to the file section
two files - the most I am concerned with is 'Mar162010'
The other is Mar172010
Mattyb
'20100316p.csv' and the wko file 'Brabender_Matthew_3_16_2010.wko'
I hope it's not just me look at things wrong.
Mattyb
starting to be convinced that xpower displayed in the summary is
including time at zero speed.
Today's ride:
Powertap cervo numbers:
Time: 3.00.46
Avg Power: 200w
Avg speed: 29.6kph
Golden Cheetah::
Time: 2.57.09
Avg Power: 196w
Avg Speed: 29.7kph
xPower: 201w
Bikescore: 183 (IF 0.744)
WKO
Time: 3.00.36
Avg Power: 200w
Avg Speed: 29.7kph
Normalized Power: 212w
TSS: 182.2 (IF 0.786)
PowerAgent (7.5.0.30)
Time: 2.57.31
Avg Power: 197w
Avg Speed: 29.13kph
Normalized Power: 211w
TSS: 201.3 (IF 0.780)
These are all with FTP set at 270W
WKO and GC performance manager chart numbers work out the same and I
know it's only a case of 10w or so, but they are all different.
Which to believe?
If I were to choose the numbers which 'feel' right, I would choose WKO
and PowerAgent.
robert carlsen
Perhaps Phil (or Andy) can illuminate, but they don't use the same algorithm.
...or we're just including zeroes.
-robert
> --
> _______________________________________________
> Golden-Cheetah-Users mailing list
> golden-che...@googlegroups.com
> http://groups.google.com/group/golden-cheetah-users?hl=en
>
> To unsubscribe from this group, send email to golden-cheetah-users+unsubscribegooglegroups.com or reply to this email with the words "REMOVE ME" as the subject.
Jamie Kimberley
of Sean and much of the GC list that zeros should be included in most
metrics even if they are due to data drops.
Jamie
"It is a mistake to think you can solve any major problems with just
potatoes."--Douglas Adams
__________________
Jamie Kimberley
Postdoctoral Fellow
Department of Mechanical Engineering
The Johns Hopkins University
Office: 410.516.5162
Mobile: 217.621.8272
Fax: 410.516.4316
E-Mail:jamie.k...@jhu.edu
Fernando Maldonado
Jim Ley
<fernando.j...@gmail.com> wrote:
> Zeros are included. I do some intervals on hills where the recovery time is
> spent coasting down the hill at zero watts. Not the way recovery should be
> done but it's what I have to work with. The results all include the time at
> zero which in my case is substantial.
This is a "feature" in WKO's implementation of Normalized Power -
Because NP is only defined where there is a reading every even time
interval (See Andrew Coggan's earlier clarification on this list) any
drop out / gap / non-recorded portion makes the equation not defined.
WKO has made the decision that time that is not recorded should not be
considered part of the duration of the activity. Like Fernando, I am
very much of a belief that this is wrong, since there is no difference
in stress between coasting down a hill at 0 watts and positive speed,
and being stopped at 0 watts with zero speed for the same duration, so
they should be treated the same.
The xPower implementation doesn't make the same decision, and I
certainly think it reflects the true cost much better.
Jim.
Mattyb
no problem with differences I've seen in the past, but I just think
something is no right here. I'm seeing over 9% difference in some
instances.
I'm not talking zero watts, I'm talking zero speed.
My first example:
Powertap cervo: avg power 259w
time: 35:50
GC
Duration: 43:27
Time riding: 34:48
Average power: 252w
xPower: 246w
WKO
time: 35:50
avg power: 259w
normalized: 271w
I've posted the files. If you have a look, you can see that there are
quite a few stops and quite a bit of time over 300w. This is not
anything close to a steady state effort. I guess I just don't
understand how such a ride can post a xPower figure lower than Avg
power for over 35mins.
Perceived effort definitely agrees with 271w normalized watts.
Another example is a ride where the time of the interval matches the
total duration instead of time riding:
Duration: 1:19:37
Time riding: 58:04
Avg watts:177
xPower: 181 vs normalized of 196w
The interval in the ride summary says duration is 1:19:37
Now I agree that zero watts should be included, but zero speed?
On Mar 21, 5:03 pm, Fernando Maldonado
<fernando.j.maldon...@gmail.com> wrote:
> Zeros are included. I do some intervals on hills where the recovery time is
> spent coasting down the hill at zero watts. Not the way recovery should be
> done but it's what I have to work with. The results all include the time at
> zero which in my case is substantial.
>
> Fernando
>
> On Sat, Mar 20, 2010 at 10:54 PM, Jamie Kimberley
> <jamie.kimber...@jhu.edu>wrote:
Mattyb
So xPower is going to include time at zero speed then.
Well I guess that explains it then.
On Mar 21, 6:01 pm, Jim Ley <jim....@gmail.com> wrote:
> On 21 March 2010 06:03, Fernando Maldonado
>
Sean Rhea
We can't pay attention to speed, as that messes up trainer workouts
where the speed sensor is on the front wheel.
Do you have RaceDay? Can you compare GC's numbers to RaceDay's? The
latter is the gold standard for xPower.
Sean
> --
> _______________________________________________
> Golden-Cheetah-Users mailing list
> golden-che...@googlegroups.com
> http://groups.google.com/group/golden-cheetah-users?hl=en
>
> To unsubscribe from this group, send email to golden-cheetah-users+unsubscribegooglegroups.com or reply to this email with the words "REMOVE ME" as the subject.
>
--
Andy Coggan
> > Zeros are included. I do some intervals on hills where the recovery time is
> > spent coasting down the hill at zero watts. Not the way recovery should be
> > done but it's what I have to work with. The results all include the time at
> > zero which in my case is substantial.
>
> This is a "feature" in WKO's implementation of Normalized Power -
> Because NP is only defined where there is a reading every even time
> interval (See Andrew Coggan's earlier clarification on this list) any
> drop out / gap / non-recorded portion makes the equation not defined.
>
> WKO has made the decision that time that is not recorded should not be
> considered part of the duration of the activity.
Not to my understanding.
Andy Coggan
Mattyb
I just downloaded RaceDay and the xPower figures aren't far off but
the average is really different.
RaceDay seems to include zeros for all figures.
Using my first example, RaceDay reported a time of 45mins or so, avg
power of 205w and xPower of 259. GC says avg of 252w and xPower of
246w.
Interesting.
I'm thinking that it should be one or the other where average and
xpower should, or not include time at zero speed. Or have a user
selection in preferences where you can choose to include zeros in your
xpower figures.
Fernando Maldonado
--
_______________________________________________
Golden-Cheetah-Users mailing list
golden-che...@googlegroups.com
http://groups.google.com/group/golden-cheetah-users?hl=en
Mattyb
more about what GC is up to (see below)
Sean, RaceDay reports completely different from GC. It looks like
RaceDay includes zeros for all figures which makes sense.
The first example:
RaceDay:
Time: 43.27
Avg Power: 205w
xPower: 258w (RI = 0.96)
Bikescore: 66.41
GC
Duration: 43:27
Time riding: 34:48
Average power: 252w
xPower: 246w (RI = 0.911)
Bikescore: 60
I guess my main concern with the above is getting avg power higher
than xpower when all other programs report differently. It looks like
GC does not include zero speed when displaying Average Power but
includes it for xPower figures.
Perhaps, as it seems to be a personal preference, there could be a
user definable option to include or not-include zeros for figures.
Just an idea, but personally I think it should be the same for both.
On Mar 21, 4:47 pm, Fernando Maldonado
<fernando.j.maldon...@gmail.com> wrote:
> XPower and Normalized Power are similar but not identical otherwise they
> would have the same name.http://www.physfarm.com/bikescore.pdfexplains it
Andy Coggan
<fernando.j.maldon...@gmail.com> wrote:
> XPower and Normalized Power are similar but not identical otherwise they
> pretty well.
>
> I also remember seeing something on the wattage forum that XPower might be
> more indicative of the effort that went into a particular activity. I'll do
> some searching and see if I can find the reference.
Phil has posted one example in which the rolling value for xPower
tracked a bit more closely with heart rate (as you would expect).
However, in developing the normalized power algorithm I found that on
average (i.e., across several hundred files) it didn't make any
difference whether you used an exponentially-weighted or a simple
rolling average to smooth the power data before weighting it.
Andy Coggan
Sean Rhea
GoldenCheetah computes xPower over the entire time of the ride. If
you stop riding, it keeps counting. That includes both points where
the power meter is reporting zero watts, *and* points where the power
meter reports no data at all. In contrast, to compute the average
power for a ride, GC just averages the power readings from all points
reported by the meter. So if the meter stops reporting, we stop
averaging in zeros. But it it keeps reporting zeros, we average those
in.
I just took a look at your Mar162010.gc file. It contains a *lot* of
stops, and given the above description of how GC works, I'm not too
surprised at the results.
To be honest, I'm not sure what the "right" way to handle a ride like
that is. If you're interested in what stress you underwent during the
ride, I think the way GC currently computes xPower is the one you
want. We're probably being too generous with average power, but
people get annoyed when we count their food stops against them.
Personally, I don't think average power is worth much for whole rides.
It's really only valuable for intervals at nearly constant power.
xPower/BikeScore/DanielsPoints are what you really want for a whole
ride, and in those cases, I think we're doing the right thing.
Speaking of which, one of the nice things about DanielsPoints is that
it doesn't matter how you handle zeros. It always gives you the same
number.
Sean
> To unsubscribe from this group, send email to golden-cheetah-users+unsubscribegooglegroups.com or reply to this email with the words "REMOVE ME" as the subject.
Eric Murray
> GoldenCheetah computes xPower over the entire time of the ride. If
> you stop riding, it keeps counting. That includes both points where
> the power meter is reporting zero watts, *and* points where the power
> meter reports no data at all. In contrast, to compute the average
> power for a ride, GC just averages the power readings from all points
> reported by the meter. So if the meter stops reporting, we stop
> averaging in zeros. But it it keeps reporting zeros, we average those
> in.
Shouldn't the two metrics (avg and xpower) be calculated using
the same data? It seems like including the no data point dead time
is incorrect for long rides. If someone takes a 45 minute stop
for lunch, that's not riding stress. But whatever the methodology,
both metrics should do it the same way.
It might be best to have check boxes for "include zeros" and
"include dead time" and use those for both metrics.
Eric
Mattyb
wanted to make sure the stress GC was indicating for a ride is
realistic.
Eric, I like the idea of having an on/off option, but, I don't have
the time at the moment to create that sort of thing.
Maybe someone else does if they want it too?
Jamie Kimberley
> Eric, I like the idea of having an on/off option, but, I don't have
> the time at the moment to create that sort of thing.
> Maybe someone else does if they want it too?
please add this as a feature request on the bug tracker. we'll be
less likely to forget it an someone may decide to tackle it on
their own.
Thanks,
Jamie
>
> --
> _______________________________________________
> Golden-Cheetah-Users mailing list
> golden-che...@googlegroups.com
> http://groups.google.com/group/golden-cheetah-users?hl=en
>
> To unsubscribe from this group, send email to golden-cheetah-users+unsubscribegooglegroups.com or reply to this email with the words "REMOVE ME" as the subject.
>
"You don't use science to show that you're right, you use science to
become right."--xkcd