Classifying passes in soccer using spatio-temporal data streams

Abstract

Over the last few years with the advent of Big Data, professional soccer teams want to build automated systems to improve decision making and reduce manual effort in post-match analysis. This work is expected to contribute towards building an automated soccer analytical system which will utilize computational geometry and data mining techniques to analyze spatio-temporal data streams to classify passes in soccer as Good, Ok or Bad ones.In this work, we describe a prototype system that preprocess spatio-temporal data and generate features which are used as input to classification functions. Features are generated based on the quantitative attributes described from the field of computational geometry with respect to soccer analysis. The created system is expected to aid players, coaches and soccer team management to analyze performance and access the quality of passes. Quality of passes has to be quantitatively evaluated to understand players’ performance and the team’s performance. Finally, this work discusses the implementation results of classification and concludes with future enhancements.