Fast Find in CPv2.0...

Terry Ferguson said:

I cannot seem to get fast find to work. I put in 2 consecutive samples at 261 Watts for the leading edge and 5 samples at 150 Watts for the trailing. I set min duration at 20 minutes and the max at 20 hours.

This should give me any portion of the ride (range of interest= entire workout) that has two data points above 261W at the beginning, is 20 minutes or more long (and less than 20 hours long= everything) and never falls below 150W for 5 data consecutive points, correct?

I can look at the data and such a range exists, yet Fast Find comes up with nothing. I thought maybe it actually needs the 5 points below 150W and that it doesn't find them because the data ends right when I crossed the finish line? Changed the last 6 points to 140W - no difference.

Did the same Find with Min Dur= 10 minutes and it finds one that it says is 19:15.42 long, yet I can find nothing near the end point that has 5 consecutive (or even 3) below 150. ?????

TF

It appears that it is ending when it finds the 5th point less than 150W - not consecutive points, just the 5th point.

TF

Terry Ferguson said:

I cannot seem to get fast find to work. I put in 2 consecutive samples at 261 Watts for the leading edge and 5 samples at 150 Watts for the trailing. I set min duration at 20 minutes and the max at 20 hours.

This should give me any portion of the ride (range of interest= entire workout) that has two data points above 261W at the beginning, is 20 minutes or more long (and less than 20 hours long= everything) and never falls below 150W for 5 data consecutive points, correct?

I can look at the data and such a range exists, yet Fast Find comes up with nothing. I thought maybe it actually needs the 5 points below 150W and that it doesn't find them because the data ends right when I crossed the finish line? Changed the last 6 points to 140W - no difference.

Did the same Find with Min Dur= 10 minutes and it finds one that it says is 19:15.42 long, yet I can find nothing near the end point that has 5 consecutive (or even 3) below 150. ?????

TF

The way I read that logic is: Find me 20+ minutes that do not drop below 261. But you only have a 19:15.42 stretch over 261. The leading edge sets the criteria for the minimum time and the trailing edge says when ot stop.

You can confirm this by seeing if any records exist below 261 for the 19:15.42 section. I bet there are not.

Perfered logic:
I would actually like to say find all section that average X Watts. With a min time. That way I can say find all 20 min itervals that average at my FTP, for example. Handy for Microbursts and rolling TTs. Or find all 3+ minute periods between bottom and top of say your VO2 max range. But again you would not find the effort that has even One record below VO2. The average for min time would be more useful IMHO.

Just a thought for the next build

jetnjeff said:

The way I read that logic is: Find me 20+ minutes that do not drop below 261. But you only have a 19:15.42 stretch over 261. The leading edge sets the criteria for the minimum time and the trailing edge says when ot stop.

You can confirm this by seeing if any records exist below 261 for the 19:15.42 section. I bet there are not.

Perfered logic:
I would actually like to say find all section that average X Watts. With a min time. That way I can say find all 20 min itervals that average at my FTP, for example. Handy for Microbursts and rolling TTs. Or find all 3+ minute periods between bottom and top of say your VO2 max range. But again you would not find the effort that has even One record below VO2. The average for min time would be more useful IMHO.

Just a thought for the next build

No, that doesn't work. There are many points below 261. Also, if that was the logic, why would you have a place to enter a trailing edge number? TF

Terry Ferguson said:

No, that doesn't work. There are many points below 261. Also, if that was the logic, why would you have a place to enter a trailing edge number? TF

Trailing edge, you are probably correct. But if you are finding a 19 min interval and not a 20 min, some how there is not a 20 mint interval that meets the criteria. I have to admit, that when I use the Fast find I have to play around and use multiple criteria to find what I want. So I usually add the Level grid lines and find them manually. I find it faster than the fast find.🙂

Six months later and I still haven't figured out how it works. I have attached a sample file (change the extension to wko).

Criteria:
Leading edge: Consecutive samples= 1 Equal to or more than 330 Watts
Trailing edge: Consecutive samples= 10 Less than 280 Watts
Minimum duration 30 sec
Maximum duration 60 min

Through 28 minutes, it works as I would expect. Then why doesn't it find the 29 min set? On the 39 min set, why does it stop when the next samples are still over the final wattage (as well as the initial wattage)? Why does it pick up the 53 min and not the 51 min?

I checked the CP forums, but all that I found were the questions.

It doesn't seem to be of much use in finding 'matches' if I don't know what its criteria really are.

TF

EDIT: Now using CPv2.1/WKO+ if that isn't evident by the file extension.

Terry Ferguson said:

It doesn't seem to be of much use in finding 'matches' if I don't know what its criteria really are.

Is that why you want to use "Fast Find," for match analysis?

I could never get fast find to produce any meaningful results, even when actively trying to 'find' an interval that I was staring right at on the screen (ie, a 4-min L5 interval done on the trainer with a distinct recovery period immediately before and after it). I gave up, but would be happy to learn that there's some trick to it.

frenchyge said:

I could never get fast find to produce any meaningful results, even when actively trying to 'find' an interval that I was staring right at on the screen (ie, a 4-min L5 interval done on the trainer with a distinct recovery period immediately before and after it). I gave up, but would be happy to learn that there's some trick to it.

One thing that baffled me for a while was the interval drop down box. It would not take the entire work out by default if any manual ranges were created.

So I would think I was searching the whole file but was only searching my first manual interval.

Figuring this out helped but there still is a better way to set perameters, Average or NP for the allotted time frame.

Although the calculation may take a while as it has to check each point set of data.

RapDaddyo said:

Is that why you want to use "Fast Find," for match analysis?

Well, when I started to analyze a race file (last spring) I thought that Fast Find 'matches' would be a good place to start. But I found that it couldn't find ones that I was sitting there looking at. I've been amazed that this hasn't been discussed and has been ignored on all the power forums (inculding CP) whenever questioned. - TF

Terry Ferguson said:

Well, when I started to analyze a race file (last spring) I thought that Fast Find 'matches' would be a good place to start. But I found that it couldn't find ones that I was sitting there looking at. I've been amazed that this hasn't been discussed and has been ignored on all the power forums (inculding CP) whenever questioned. - TF

I went to a conference in March and Hunter skimmed over and said you have to play with it.

I figured that they would work on this for future builds.

Terry Ferguson said:

Well, when I started to analyze a race file (last spring) I thought that Fast Find 'matches' would be a good place to start. But I found that it couldn't find ones that I was sitting there looking at. I've been amazed that this hasn't been discussed and has been ignored on all the power forums (inculding CP) whenever questioned. - TF

FWIW, I've ignored it because I'm ignorant on the topic: I've never used the Fast Find feature, so can't really say anything about how/how well it works.

It's definitely not a perfect algorithm, but at the same time, I haven't found a file I couldn't get it to work on. It really just takes an understanding of what you are looking at and how to define the 'matches'. Mind you, if every single 'match' you are looking for has 5 samples under the trailing edge power, and you have 5 consecutive samples selected, then you'll never get the algorithim to work. You really do have to play with it. Once you understand it and use it, it's a nice tool.

Sorry, there is no hard and fast rule to use it. I gave a presentation at the USA Coaches Summit on Friday last week and picked out a random race, and started doing some fast finds on it. The first 2 came up empty, but those 'mistakes' allowed me to better define the criteria and I got them nailed each time afterward.

Hope this helps.

Hunter

Terry Ferguson said:

Well, when I started to analyze a race file (last spring) I thought that Fast Find 'matches' would be a good place to start. But I found that it couldn't find ones that I was sitting there looking at. I've been amazed that this hasn't been discussed and has been ignored on all the power forums (inculding CP) whenever questioned. - TF

Hunter w/kg said:

It's definitely not a perfect algorithm, but at the same time, I haven't found a file I couldn't get it to work on. It really just takes an understanding of what you are looking at and how to define the 'matches'. Mind you, if every single 'match' you are looking for has 5 samples under the trailing edge power, and you have 5 consecutive samples selected, then you'll never get the algorithim to work. You really do have to play with it. Once you understand it and use it, it's a nice tool.

Sorry, there is no hard and fast rule to use it. I gave a presentation at the USA Coaches Summit on Friday last week and picked out a random race, and started doing some fast finds on it. The first 2 came up empty, but those 'mistakes' allowed me to better define the criteria and I got them nailed each time afterward.

Hope this helps.

Hunter

"Hope this helps." Sorry, but no.

"Once you understand it and use it, it's a nice tool." That's the problem - I cannot understand it. What is the algorithm? I cannot know the value of what it finds if I don't know what it looks for. In the sample file I attached, I can find no pattern. Sometimes it stops on a sample that is over the trailing value when there are many more samples over that value after it???

TF