Restaurant Growth Potential Modeling and Strategic Classification: A Data Science Framework for SkyCity Auckland Restaurants & Bars

Expansion decisions across restaurant portfolios are often uniform and intuition-based, even though branches differ sharply in cost structure, channel dependence and delivery economics. This paper presents a framework for classifying and scoring 1,696 branches of SkyCity Auckland Restaurants & Bars for growth readiness.The pipeline engineers 19 features from a 30-field operational dataset, corrects skew and scale, reduces dimensionality with Principal Component Analysis (6 components, 90.5% of variance), and groups branches into six strategic archetypes using K-Means, validated against Ward hierarchical clustering, DBSCAN and a ten-seed stability test. A transparent Growth Potential Index (GPI, 0-100) combines growth signal, cost resilience, channel balance and logistics scalability, and a rule layer assigns each branch one of three actions: Optimize, Rebalance channels, or Hold / Stabilize. A weight-sensitivity analysis shows the rankings are stable under alternative weightings.Key findings: Uber Eats and DoorDash earn roughly 0.8% net margin against 26.8% in store; platform commissions are about 1.8 times the portfolio’s total monthly net profit; and all 408 loss-making branches are full-service restaurants. Limitations include a single monthly snapshot and expert-set index weights.The accompanying code, scored dataset and Streamlit dashboard are described in the companion item. Live dashboard: https://unified-mentor-project-02-f552bweqzd8dischphjdvz.streamlit.app/ This work was prepared as a deliverable of the Machine Learning Internship at Unified Mentor.