# Help from the TR community for my school project (Results)

hello fellow TR users, I would like to ask you for a little help with my uni course. I have to do some kind of “project” for one of my university courses and luckily I can choose my own data. Therefore I would like to connect it with something I love Please answer these 7 really short questions. It would be really helpful!

3 Likes

Done, good luck!

Done, hope it helps!

For the triathletes among us do you just want bike training info?

Done!

Done…but for km ridden in last year…well nearly zilch as all training done on TR so really apart from races and v occasional ride outdoors then mileage in zero (don’t count or record mileage indoors as pointless random number!)

I agree that it is a little bit misleading, because a lot of training is made indoors… I had more ideas on possible factors such as average IF during week, time spent in zones upper than 4…, but I tried to make it simple, in hope of more answers. I definitely think that something interesting could be made with such numbers… I suspect that the TR team is doing something, there were some rumors about AI models and I would love to see that in action and also the “big” data of TR

No worries, just wanted you to be aware of anomalies in your data if got someone saying they train 5-6 hours every week and yet annual mileage is about 100km!!!

let us know the results!

Done. No clue on annual distance. Absolutely no clue. Sorry.

Done!
Good luck with your project and study!

Oh, btw… What is this research about? I’m honestly curious. And will (and can) you share a summary or your endresult?

I also had to guesstimate annual miles. I might have underestimated but hopefully not to much…

I am trying to make some clustering of my data and divide the cyclists into groups that are “similar”

So far, interestingly, the mileage of last year and FTP intersect is well divided and high mileage people are in the highest FTP group. Interestingly if we look on w/kg (FTP/kg) and age intersect the best group seems to be formed of the youngest cyclists and cyclists in the middle ages. I always had a theory that we “amateurs” are best when we do not have kids yet or we are in the ages when the kids are not that young anymore. Also weekly training hours and FTP or w/kg are well divided into groups, I think that there are more things that could be justified by our common sense of people in the sport and are also clustered in that way… Well I hope that I would get even more data and something meaningful would come out and if not it was a good stats lecture

5 Likes

Done interested to see the results too

Firstly thanks again to everyone who has participated. I got A and the second theoretical question was not even needed, we have talked too much about cycling and instead of 50 minutes, I was there 1:05
So the results:
Firstly a simple summary table:

Secondly, the clustering (I had to standardize the data, but you can compare what is the “zero” with the summary table where you can find the mean. Moreover the standardized data are better to visualize ):

By the way, more training and lower weight make one cluster and low weight and high FTP another, so train smart
And lastly a 2D plot of every characteristic (note that 3D plot would be better but its hard to post a picture of it so it would be readable… ) The lines shows correlations so for example distance ridden in the last year and weight in kilos is negatively correlated - more km -> lower weight and vice versa. Therefore the left side is the best group.

You can identify yourself according to this table:

And lastly, this is how it looks if you plot the original data:

I hope that this post would be an interesting addition to you coffee/tea routine and thanks again I did not write long interpretations of the groups as in my project`s PDF, I am sure that every cyclist/triathlete would see it instantly.

4 Likes

Great job! Glad you got the A. Thanks for sharing