This study proposes a network analysis of the complete set of table tennis matches played in a Country. A wide dataset concerning all results registered by the Italian Table Tennis Federation (FITET) is considered. The dataset includes about 1 million matches played by 26k tennis table players in last 8 seasons. In the resulting directed and weighted network, players are represented by nodes and matches by arcs between nodes. We investigate the main features of the network, which shows scale-free and small-world properties, typical of a "complex network". The scale-free degree distribution indicates that few players played a relatively high number of matches, while most players with few matches are in the long-tail. In addition, most nodes are connected by very few edges. Assortative mixing is positive, to indicate that players often play with other players having similar degree. A second analysis focuses on the correlation between the improvement in ranking position and centrality measures, in one season. The favourable position of an athlete in the network, as identified by betweenness, closeness, eigenvector and degree centrality is positively correlated with the improvement in the final ranking. This underlines the importance of the adversaries and the kind of tournament an athlete is playing (i.e. young or adult adversaries, specific tournaments or Championship). Finally, we investigate the main topological features by applying community detection algorithms, which help in detecting groups and communities. Visualisation allows to clearly identify these features and structural patterns.
© Copyright 2017 Proceedings Book of The 15th ITTF Sports Science Congress Düsseldorf, 27th - 28th May 2017. Veröffentlicht von International Table Tennis Federation. Alle Rechte vorbehalten.