STAR | Devpost

SAP Challenge
Federato Challenge
Touch Bistro Challenge

📖 About the Project

🔥 Inspiration

The CxC UW 2025 Datathon provided an exciting opportunity to apply data science and machine learning to solve real-world challenges. Our team was driven by the prospect of analyzing large datasets, uncovering insights, and proposing data-driven solutions for businesses and global issues.

🏆 Accomplishments We’re Proud Of

✅ Developed a Multidimensional Poverty Index (MPI) that ranks countries based on education, health, and living standards.
✅ Built accurate forecasting models (Prophet & Exponential Smoothing) to predict restaurant sales trends.
✅ Created a Markov Chain & Graph-Based Model for real-time next-action recommendations to enhance user engagement.
✅ Implemented PCA & Random Forest feature selection, identifying the most impactful indicators for poverty and restaurant analytics.
✅ Designed data-driven policy recommendations based on clustering analysis and causal inference techniques.
✅ Overcame messy data challenges and fine-tuned machine learning models for optimal performance under time constraints.

🎯 What We Learned

Through this project, we gained valuable experience in:

Data Preprocessing & Cleaning – Handling missing values, standardizing data, and feature engineering.
Exploratory Data Analysis (EDA) – Identifying patterns in complex datasets.
Machine Learning & Forecasting – Applying PCA, Clustering, Time-Series Analysis, and Graph Analytics.
Causal Inference & Predictive Modeling – Understanding feature importance and next-action recommendations.
Business & Policy Insights – Translating analytical results into meaningful recommendations.

🚀 What's Next?

We aim to extend our work and explore:

🏗 Deploying a real-time recommendation system for engagement optimization.
📈 Enhancing forecasting models with deep learning techniques (LSTMs, Transformers).
🌍 Applying our MPI framework to broader datasets for global poverty insights.
🤝 Collaborating with industry partners to implement our findings in real-world settings.
🎨 Developing interactive visualizations for more user-friendly data interpretation.

Despite the challenges, this experience deepened our understanding of real-world data problems and reinforced our passion for AI, data science, and problem-solving.