CP question about MMP graph

msummers said:

When you change the Mean Maximal Power Graph to read normalized power, does that mean that the times displayed on the curve (dragging the cursor) are the best NP times for that time frame?

Correct.

msummers said:

I take it that this is not be available anywhere else (best NP power for a given time).

I believe another way to see best NP for different durations would be to create a custom periodic chart where best NP is charted, and then select the duration of interest (ie, 1m, 5m, 20m) in the chart options.

msummers said:

I would then also assume that it would not be possible to see the power graphed presentation of that best NP for a given time frame, because the other data chart on the graph tab can only sort by best AP (average power).

I didn't understand your question here.

msummers said:

Also does the power distribution chart use NP or AP?

It uses instantaneous power, and totals the number of data points that fell into each bin. It's raw data with no averaging or normalizing involved.

[quote="frenchyge"]Correct.



I didn't understand your question here.


Thanks Frenchy.

My question was about if I identified a CP20(NP)value on the MMP curve, could I find that time period on the graph tab where your best 20min time is based on AP only? I'm talking about the graph on the ride file.

ms

msummers said:

I identified a CP20(NP)value on the MMP curve, could I find that time period on the graph tab where your best 20min time is based on AP only? I'm talking about the graph on the ride file.

Unfortunately, no.

acoggan said:

Unfortunately, no.

Andy
Could it then be recommended that whenever the Variability Index is high(> 1.5) then the MMP graph NP values are a better ( or at least a good ) performance analysis for an individual ride. That is as opposed to the best AP values on the ride power graph. I could see where if you try to look for some performance gains in different durations over a combination of flat and high VI rides, things would be hard to interpret.

ms

msummers said:

My question was about if I identified a CP20(NP)value on the MMP curve, could I find that time period on the graph tab where your best 20min time is based on AP only? I'm talking about the graph on the ride file.

Actually, I think the answer to your question is yes. Subject to correction by Andy, here's what I think is going on with the MMP curves. The system searches the files based on AP. The MMP NP curve is based on the NPs for the same durations in the same ride files. Now, in theory, the max 20min NP duration is not identical to the max 20min AP duration. Put the two curves (MMP AP and MMP NP) side by side and cherry pick a few durations on the AP curve. Then, look at the NP curve and you will find the identical durations in the identical ride files. Coincidence? I don't think so. I think it is searching based on AP only. One of these days, I'm going to take a bunch of ride files and find the actual max NP durations, just for fun.😉

RapDaddyo said:

Actually, I think the answer to your question is yes. Subject to correction by Andy, here's what I think is going on with the MMP curves. The system searches the files based on AP. The MMP NP curve is based on the NPs for the same durations in the same ride files. Now, in theory, the max 20min NP duration is not identical to the max 20min AP duration. Put the two curves (MMP AP and MMP NP) side by side and cherry pick a few durations on the AP curve. Then, look at the NP curve and you will find the identical durations in the identical ride files. Coincidence? I don't think so. I think it is searching based on AP only. One of these days, I'm going to take a bunch of ride files and find the actual max NP durations, just for fun.😉

I've looked at my MMP curve (AP) and then switched to NP and found that the Max NP instance for a given duration actually happened on a completely different day than the Max AP instance. I think the MMP does a completely new search when you switch from AP to NP.

frenchyge said:

I've looked at my MMP curve (AP) and then switched to NP and found that the Max NP instance for a given duration actually happened on a completely different day than the Max AP instance. I think the MMP does a completely new search when you switch from AP to NP.

Okay, after more careful examination I have found a few instances where the ride files are different at the same durations. But, answer this. When I dump the two curves as raw data, the duration data points are identical in both curves. Coincidental?

frenchyge said:

I've looked at my MMP curve (AP) and then switched to NP and found that the Max NP instance for a given duration actually happened on a completely different day than the Max AP instance. I think the MMP does a completely new search when you switch from AP to NP.

Okay, I've looked into it a little further and here's what I think is going on. Within a ride file, the max NP is not searched for independently of max AP. Rather, max APs are identified and NP is computed for each duration. Then, in the MMP curve, it searches the max NPs, which are in fact associated with the max APs and not computed independently. So, while the ride files can be different, the inherent underlying computation is not based on an independent identification of max NP. I'm going to rip into a ride file and figure this out when I get a few minutes. Now, it's bugging me.

RapDaddyo said:

Okay, I've looked into it a little further and here's what I think is going on. Within a ride file, the max NP is not searched for independently of max AP. Rather, max APs are identified and NP is computed for each duration. Then, in the MMP curve, it searches the max NPs, which are in fact associated with the max APs and not computed independently. So, while the ride files can be different, the inherent underlying computation is not based on an independent identification of max NP. I'm going to rip into a ride file and figure this out when I get a few minutes. Now, it's bugging me.

I still don't think that's the case. Go out and do a few sprints with ~2 minute recoveries in between, and then do a 10 min FT interval. I'm pretty sure you'll find that the max NP value occurs during the sprint interval portion, and the max AP correlates with your FT interval.

I'm pretty sure I've looked at the NP MMP curve and noticed a particular hump that was higher than I thought I'd ever done for that duration. Looking at the ride file, it had picked up a high NP from a portion of the ride that had a pretty low AP, and thus had escaped my notice.

frenchyge said:

I still don't think that's the case. Go out and do a few sprints with ~2 minute recoveries in between, and then do a 10 min FT interval. I'm pretty sure you'll find that the max NP value occurs during the sprint interval portion, and the max AP correlates with your FT interval.

I'm pretty sure I've looked at the NP MMP curve and noticed a particular hump that was higher than I thought I'd ever done for that duration. Looking at the ride file, it had picked up a high NP from a portion of the ride that had a pretty low AP, and thus had escaped my notice.

I stand corrected. I created a "trick" file to create a major discrepancy between the ride's max 10min AP and max 10min NP. Although the program was off by 3W (on the max NP segment), it found both max durations independently. I still can't figure out why the durations are identical between the MMP AP and NP curves.😕 Maybe it just has pre-defined cut points.

Hope this is not too off-thread.

Is there a known problem with the "Copy Raw" function for the mean maximal power chart? The numbers I get when I do this are not the same as those which appear in the chart.

As an example a chart which starts with 900W at 1sec, 280@45sec, 240@10min etc. produces Raw numbers which start [email protected]@45sec... etc.

JitensyaJim said:

Hope this is not too off-thread.

Is there a known problem with the "Copy Raw" function for the mean maximal power chart? The numbers I get when I do this are not the same as those which appear in the chart.

As an example a chart which starts with 900W at 1sec, 280@45sec, 240@10min etc. produces Raw numbers which start [email protected]@45sec... etc.

That is perplexing. I get the same thing. I can't remember what the magic decoder ring is (because I never use the Copy Raw function), but I think it is in a thread around here somewhere.

RapDaddyo said:

That is perplexing. I get the same thing. I can't remember what the magic decoder ring is (because I never use the Copy Raw function), but I think it is in a thread around here somewhere.

Thanks. Glad it's not just me. What do you mean by decoder ring? Do you mean that the numbers are all there, they just need modifying somehow? At first I thought maybe there was a problem with having the graph in log scale, but that seems not to affect things. At second glance I thought that the first few values (short times) were missing, but it's more complicated than that.

What I initially wanted to do was to compare, for example, a curve for this month with one from last month, on the same graph. Doesn't seem possible with CP as it is, so I thought I could get the raw numbers out and plot them myself. Do you have a better way of doing things like this?

I'm sure it's of not much interest to most, but this shows a comparison between CP's graph(left) and the results of "Copy Raw" (right) for a particular case. Scales are roughly the same.

[img]homepage2.nifty.comMMP compare.gifOpen ↗[/img]

JitensyaJim said:

Thanks. Glad it's not just me. What do you mean by decoder ring?

I mean an answer to your question. It came up before, but I'm not sure in which thread. I seem to recall that there was an answer.

JitensyaJim said:

What I initially wanted to do was to compare, for example, a curve for this month with one from last month, on the same graph. Doesn't seem possible with CP as it is, so I thought I could get the raw numbers out and plot them myself. Do you have a better way of doing things like this?

Well, if what you want to do is compare the same chart for two different time periods, you can clone the chart and change the dates. Likewise, if you want to compare the same chart on an AP vs. NP basis, you can put the charts side by side. If you want to extrapolate the data and do something with it outside of CP (e.g., in Excel), I'm not sure how to get the duplicate data.

I don't use those charts much unless I want to cherry pick a couple of values, in which case I just use the CP chart as is. My needs outside of CP relate to ride file parsing by training level, for which I have written my own program to act on the raw ride files. My other ex-CP need is the Critical Power Curve and again I just use a few selected numbers which I copy manually. So, I almost never need to use the Copy Raw function for the charts.

RapDaddyo said:

I mean an answer to your question. It came up before, but I'm not sure in which thread. I seem to recall that there was an answer.

Thanks -- I'll have a hunt around.

Also thanks for the suggestions. I had got as far as having several cloned charts of various types, just thought it would be nice to have them all on the same graph.

Edit 10 mins later.......
Just found something on the CP forums

http://www2.cyclingpeakssoftware.com/forums/viewtopic.php?t=341&highlight=copy+raw

It seems that in this case "copy raw" only "copies raw" for the latest ride and not neccessarily for the data actually on the chart. The CP thread refers to build 82, mine is 109. Perhaps the bug is fixed in "WKO+"?