Since the player-tracking data for basketball became available, researchers have tried to figure out the best way to utilize it. These data are rich and spatio-temporal, and previous work have attempted to extract knowledge from them using advanced statistics, feature engineering, supervised machine learning, and other methods that are commonly employed to handle data of this nature.
In our work, we hope to use supervised machine learning models in order to understand the relationship between player movement on offense during a possession and concrete successful outcomes such as quality shots. However, instead of inserting any preconceived basketball knowledge with feature engineering, we hope to utilize deep learning models such as CNNs to automatically discover features about player movement that relate to offensive success. These models could then provide insight into offensive strategy in basketball or even be the basis of quantitatively-based team or player evaluation metrics.
If you would like to contact us about our work, please scroll down to the people section and click on one of the group leads' people pages, where you can reach out to them directly.