Forecasting extremes of football players' performance in matches

This repository contains material auxiliary to paper titled "Forecasting extremes of football players' performance in matches", divided into the following parts.

A. Athlete performance parameters, as generated from the Apex Pro Series, STATSports, Premium System607
2023, Sonra 4.0

• "Appendix A.pdf"

B. Sample datasets used in modeling

• "Appendix B.xlsx" - an athelete's performance log as produced by STATSports system
• match-day-gps.csv - an athlete's GPS trace on a match day (MD)
• md-5-training-gps.csv - the athlete's GPS trace on a training session 5 days before MD

C. Sample models and modeling results

• Z_cnt.py - sample pre-processed and aggregated GPS data, used further in modeling
• zenodo_gps_demo.py - a scipt examining a range of time vs. speed definitions of an interval that correlate training and match performace best
• zenodo_apx_demo.py - a script generating sample models from APX-Data