TrainerRoad's Big Data


#58

This is a really interesting topic! When discussing the FTP distribution on AACC, I have Chad and Nate mention analyzing factors such as intensity reduction and work out failures. I guess that is one more external motivator for me to complete the workouts besides the pic on Strava:wink:.

Since I trust TR to do something awesome with the collected data, I want to send them the best data possible. This thought prompted two questions on my part:

  1. If I turned off erg mode, but left the intensity alone, how would this be interpreted? Would it be a successful workout so long as the TSS and NP were in range with the goals?

  2. Workouts emulated outside or on other platforms but pulled in via Strava, are these contributing in any way?


#59

I would guess your first question would be the same results of those not using a smart trainer. At what point it’s considered a fail is a good question though. Is it by dropping a certain amount of percentage or stopping the workout early?


#60

Very good point. Cycling is already an awesome practical application of IoT (Internet of Things), and data & analytics, big data, etc, usually follow shortly behind IoT. Semi artificial intelligence could go a long way. And this has been my complaint with most all software for cycling data out there, even GoldenCheetah and WKO4, which is that they produce all this nice data, charts after charts, table after table, trend line after trend line, but what do we do with the data? It’s pointless if it doesn’t incite us to action and changes in strategy, The point of having big data is to dive into data based decision making, more importantly, strategic data based decision making – decisions that change our training plans, approaches, etc. I had to do this myself in the case of WKO4. I took my power curve, which showed a dip in 20-40s (i,e., limiter), and I modified my entire training plan to remove that dip. That is just one very simple example. Imagine the potential


#61

I think @occasionalathlete came up with a definitive list early on in this thread.

The key points to me are personalised analysis of cadence and how it correlates to power and heartrate.

Dynamic suggestions on current FTYP - similar to WKO4 mftp calculation

Given an objective, crit, TT climb on X date what plan/sessions are best suited towards that


#62

Maybe I’m thinking small, but I would start with:

  • Personalized ATL & CTL exponential decay rate
  • Personalized work to recovery weeks. That is, if I’m doing SSBMV1, TrainerRoad calculates that I can handle 4 weeks of work before I need a recovery week, and modifies the plan accordingly. But when I’m doing Short Power Build MV1 (SPBMV1) I can only handle 3 weeks of work before needing a recovery week, and TrainerRoad modifies SPBMV1 accordingly

#63

The dream is to get people to do the ramp test once and never have to do it again as long as they keep training. If you take a hiatus from training we’d need you to ramp test again.

We’re working on all the blocks to make this happen.


Poll: Would you ramp test if you didn't need to?
#64

I agree.


#65

We’re still playing with this. I don’t want to give too much away until it’s launched because: 1) We might change it. 2) Competitors could read it and start copying us.

:smiley:

We’re working on all of these cases.


#66

And also why stopped early - sometimes I need to cut workout 10-15 mins short (or choose say 75min version instead of planned 90 min) due to life commitments rather than physically unable to complete workout…


#67

Marvelous!


#68

I’d be interested to know how you see this working in broad strokes (no need to give away trade secrets).

Here’s my take on some difficulties that you might face:

Take the WKO4 solution which plots the Power Duration curve based on the rider’s Mean Maximal Power curve and then takes a turn-point on that line to be indicative of FTP. The model is fairly robust as long as you have enough max efforts.

For that type of model one of the problems I can see is that if you’re just training indoors (as I imagine a high percentage of folk do during the winter), or only have power data for indoor rides (do you know what percentages of users own a power meter?) then the PD curve is self-defining over the course of a plan. There would be nothing that you could tell about the riders abilities after the completion of the plan that you couldn’t have known before they started, assuming everything was completed as prescribed.

Without outdoor rides or tests there would be no max efforts since the intervals are not tests of MMP: My own ability to crush an 8 minute test far outweighs my ability to perform 4 x 8 with 3 minutes rest.

There is also the discrepancy in power between indoor and outdoor rides: it wouldn’t be appropriate to base indoor workouts on a model driven from maximal efforts done outside.

Obviously this is only one way of modelling FTP - there may well be others out there but I’m pretty sure they all require maximal efforts (tests) at different durations.

It’s an interesting topic and was actually the one that I had in mind when I first started the topic. Having said that, it is actually only the determination of the model that requires Big Data, the calculation of FTP only needs a single user’s data set.

Mike


#69

Just have workouts that have the last interval be a ‘modeled fail’. If their numbers say you should be able to hold 105% ftp for 5 min and they build an interval that tells you to hold it for 7 min and do… then they know your FTP is higher, and could adjust accordingly.


#70

How would you overcome that repeatable interval power is never on the PD curve? If it’s just a single interval you might as well call it a test.

Mike


#71

Because they aren’t saying you can maximally hold this power for this time. They are saying “people that can hold this power for this time at the end of this workout will statistically be able to complete these workouts at this intensity.”

Take everyone’s favorite SSBIMV O/U progression; Reinstein, Tunemah, McAdie, Palisade, McAdie +1. When you finish McAdie +1 they can probably make a pretty good guess at the power that it would take to make you fail Reinstein.


#72

A good answer.

How you develop a model based on that principle may be difficult though as it would only be the last interval or interval set that would be done to to exhaustion. Is there even any data out there that would allow us to build that model?

Mike


#73

Their data.


#74

But these intervals don’t really exist yet because what you are proposing is a change to the current workouts.

Don’t get me wrong - think there could be the start of a good model but not sure that there is enough verified data to build it with.

Mike


#75

There are some good ideas in here and some of them hit pretty close to what we’re going to do. I can’t share everything as some stuff is a competitive advantage if we can execute.

I just wanted to say ya’ll rock :smiley: :metal:.


#76

Interesting thoughts from @GPLama & crew.


#77

…and here as well. TR-specific mention