Author(s): MALLIARAS GAVRIELATOS, T., WARMENHOVEN, J., COUTTS, A., SHEEHAN, W., Institution: UNIVERSITY OF TECHNOLOGY SYDNEY, Country: AUSTRALIA, Abstract-ID: 1199

INTRODUCTION:
Within the golf literature there is a discussion on how developments in golf science have not yet had as significant an impact on coaching and training practices as they should. This may, in part, be due to the absence of an appropriate theoretical framework for the empirical analysis of the golf swing. The objective of this scoping review is to examine the current statistical and AI methods that are used in the analysis of golf swing data, made available from launch monitors.
METHODS:
A scoping review was conducted using SCOPUS, SPORTDiscus, Inspec, and ACM, to identify studies up to August 2025. A total of 29 articles were identified and five met the inclusion criteria. Forward and reverse citation searches were also conducted on these articles, identifying an additional 123 records. Of these, eight met the inclusion criteria, resulting in a total of 13 papers for data extraction. Data extraction from these articles revealed specific characteristics regarding the use of different statistical paradigms as well as identifying particular methods that are currently favoured in the analysis of golf swing and performance data, collected by launch monitors.
RESULTS:
The prediction paradigm was the most frequently utilised. Certain articles also conducted inference to discern particular attributes that differ between skilled and non-skilled golfers. A select number of studies used a mix of parametric and non-parametric methods. While these studies facilitate an understanding of which launch monitor variables are most highly correlated with performance outcomes, and verify differences between different skill levels, they do not make use of the launch monitor data to inform coaching and training practices.
CONCLUSION:
This review demonstrates how statistical and AI methods are currently being used in the analysis of golf swing data. However, there is no discernible, standard framework for which this analysis is conducted. Due to the prediction paradigm being favoured, no insights into improvement of performance outcomes can truly be made. Future research should focus on implementing methods from an inference paradigm, to better inform coaching and training practices in golf.