Principal Component Analysis in the Study of Structure of the Best Polish Decathlon Competitors from the Period between 1985–2015

Abstract

The modern decathlon is a sport consisting of ten different events held over two days, played by men. Depending on the complexity of combined events, variety of events (runs, throws, jumps), the multi-stage, time-consuming and difficult training process the sport is considered as one of the most difficult. The analysis of careers of the best decathlon participants and applying advanced data-mining methods can help define the patterns occurring between each decathlon event and the final result. The research material encompasses career data of the 25 top competitors from Poland in years 1985–2015. Principal component analysis (PCA) was used in the research in order to designate new uncorrelated variables (components), representing input data across a new plane. Data analysis involved appointment of correlations between the events, determining the number of main components taken into account in further studies, analysis of the weight of each variable in formation of main components as well as visualisation and interpretation of results in the new plane described by the determined main components. Through the implementation of PCA method in the process of analysis it was possible to designate over 69% of compound data volatility with the use of the first three components. The first component, comprised of seven variables, displays the largest share in the total variability. The study of the relationship between variables in the new plane displayed strong correlations between sprint events (100 m, 110 m hurdles) and long jump and pole vault. No correlations between the 1500 m run and other events were found.

Keywords

• decathlon
• sport career
• principal component analysis