Ultimate Performance Analysis Tool (uPATO)

This chapter introduces the concept of network analysis and presents a series of software tools developed with the objective of analyzing networks. This analysis is usually performed on a set of metrics calculated on the network or on a representation of the network. Most of the existing tools are not free or are limited to unweighted networks, ignoring, in the case of weighted networks, the weight of the edges that connect the nodes. For example, in team sports, we have weighted networks, where edges represent the interactions between team players and their weight its frequency.

This chapter presents an overview of uPATO application, which was designed mainly for network analysis applied to team sports. However, this tool can be used for any network that can be represented by an adjacency matrix (e.g., a computer network, a telecommunication network, or even a social network, etc.). Thus, a first module was developed to allow codify the network, which, in the case of team sports, is given by a matrix with the sequences of interactions between teammates (i.e., a digraph). But this tool was designed to support graphs and digraphs, weighted and unweighted. Team sports are a good example of the necessity for calculating metrics on weighted networks. uPATO was developed with the main objective of analyzing team sports, where weights represent the frequency of the interactions between players, providing fundamental information on the analysis of the team factor. It calculates metrics in both weighted and unweighted networks and separates metrics into three major categories: individual metrics, subgroup metrics, and team metrics. Beyond metrics that uPATO allows calculating, a representation module that allows visualizing the network was also developed (i.e., the digraph or graph, weighted or not) and some charts for the data were calculated. Besides, the uPATO tool has an additional module for processing geolocation data. Currently, some teams use GPS devices to have the position of the players during the match (e.g., FieldWiz and TraXports formats are supported). Thus, uPATO has a set of metrics based on geolocation data of the players. This new functionality extends the uPATO capacities for team sports analysis but also for other activities where GPS data is available. But, it does not consider yet the possibility of the ball with a GPS device. However, this additional module is out of the scope of this book but the metrics implemented are described in a previous publication [3].

This chapter contains a set of individual metrics that can be used to analyze the importance of each player in a team sport. The metrics were divided into two main categories: Centrality (Sect. 3.1) and Prestige (Sect. 3.2). Each metric includes a description of a possible interpretation of the metric, and the pseudocode to implement it. Each pseudocode describes the cases (unweighted graphs, unweighted digraphs, weighted graphs, or weighted digraphs) for which it can be used. When the description simply contains graph (or graphs), without any other specifier, it means that the pseudocode is valid for any of the four types of graphs. The included interpretation considers that the connections between the players are the passes performed between them.

A collection of metrics is presented in this chapter. This collection is categorized as meso-level (subgroup) or team metrics, depending on the scope. Each metric includes a description of a possible interpretation of the metric, and the pseudocode to implement it. Each pseudocode describes the cases for which it can be used (unweighted graphs, unweighted digraphs, weighted graphs, or weighted digraphs). When the description simply contains graph (or graphs), without any other specifier, it means that the pseudocode is valid for any of the four types of graphs. The included interpretation considers that the connections between the players are the passes performed between them.

This chapter introduces an analysis performed on real data. All described metrics were calculated for both teams and representations were created, using the Web application uPATO. The conclusions obtained from the analysis are then presented.