Oscar Richmond - Eagle Analysis

Introduction

• GPS movement data collected every few seconds
  
• Key movement variables: speed, altitude, turning angle, vertical rate
  
• Goal: uncover structure in movement behavior using PCA
  
• PCA summarizes high-dimensional movement patterns into principal components

PCA Methods

• Variables included in PCA:
  KPH, Sn, AGL, abs_angle, VerticalRate, absVR
  
• Data standardized before PCA
  
• Principal Components capture major axes of movement variation

Scree Plot

PCA Variable Plot

PCA Scores (PC1 vs PC2)

Transition to Clustering

• PCA showed clear variation in movement based on speed, turning, and vertical motion.
• The spread of points in PCA space suggests the presence of distinct movement patterns.
• PCA explains structure, but clustering helps identify actual behavioral groups.
• Next, I apply clustering methods to define and interpret these movement states.

Bootstrapped Silhouette Curve for K = 2–7

Bootstrap WSS Curve Plot

PCA Plot w/ Cluster Labels

How Movement Variables Differ Across Clusters