Basketball Movement Clustering – Eclipse Analytics Submission

Dhruv Sood A18321957 d2sood@ucsd.edu

This is my submission for the Eclipse Basketball Consulting Challenge.

What I Did

I worked with SportVU tracking data from the Raptors vs Hornets game (Jan 1, 2016).
The dataset tracks every player and the ball at 25 frames per second.

I pulled a few possessions, took the player positions frame-by-frame, and used KMeans clustering to group different types of movement. Then I added the ball's path on top in red to give context.

What the Plot Shows

• Clusters of player movement — likely showing roles like cutters, perimeter players, or help defense
• The red line is the ball's movement during the possession
• Helps show spacing and how players move relative to the ball

File

• player_movement_clustering.ipynb – code + plot + explanation all in one

Notes

Focused on making the visualization clean and intuitive. Code is in Colab and uses Python basics (no advanced packages).