Hi All,

I recently found this website / forum and it contains an mind blowing amount of information about data mining. I have a specific dataset containing the information below. Generally speaking an athlete does a series of training workouts defined as "workout type" and then participates in a event which either has a "postive" or "negative" result.

I would like to analyze the sequential patterns of training workouts which have the best result / outcome in an event.

a - workout type 1
b - workout type 2
c - workout type 3
.. - workout n

e - event 1 positive results
f - event 1 negative results
g - event 2 positive results
h - event 2 negative results
.. - event n positive/negative results

sequence examples

0 - {a,a,b,b,e}
1 - {a,b,c,f}
2 - {a,a,c,f}
3 - {b,a,a,e}
4 - {a,b,a,f,a,b,a,g}
......

The sequence will always end in a event, there could be multiple workouts/events in a pattern.

Q: What type of pattern recognition model can review the data set and indicate which patterns generally have the best results.

Q: The above doesn't take into consideration the timing between workouts and events (e.g. 7 days between workouts, or 14 days) How could that be layered on, does this change the model used?

Thanks for all who take a look at the above!



Edited 1 time(s). Last edit at 05/30/2019 09:25AM by ccmd00d.

Hi,

That is a very interesting topic. If you want, you can send me an e-mail to philfv8 AT yahoo.com

I can share some ideas with you about this and perhaps some unpublished paper that would be relevant for this.

Best regards,

Philippe