About & Contact

Learn more about the author and get in touch

About This Project

This portfolio project showcases a comprehensive machine learning analysis of Airbnb listing data from New York City. Originally completed as part of CPSC 330: Applied Machine Learning, the analysis demonstrates end-to-end ML workflow from data exploration to model deployment.

The project predicts listing popularity using reviews per month as a proxy metric, achieving an impressive R² of 0.6956 on the test set. This website transforms the original Jupyter notebook into an interactive, portfolio-ready presentation suitable for both technical and non-technical audiences.

Technical Stack

Analysis: Python, scikit-learn, pandas, matplotlib, seaborn, LightGBM, SHAP
Website: Next.js, TypeScript, TailwindCSS, Recharts, Vercel

Key Achievements

Academic Excellence

  • • Comprehensive ML workflow implementation
  • • Advanced feature engineering techniques
  • • Model interpretation with SHAP
  • • Statistical validation and testing

Professional Skills

  • • Full-stack web development
  • • Interactive data visualization
  • • Responsive design principles
  • • Modern deployment practices

Benjamin Gerochi

Data Science & Machine Learning Enthusiast

Project Resources

Download Original Notebook

Complete analysis with code, visualizations, and detailed explanations

Project Methodology

1. Data Exploration

Comprehensive analysis of 48,895 NYC Airbnb listings with 16 features.

  • • Missing value pattern analysis
  • • Feature distribution examination
  • • Correlation analysis
  • • Geographic pattern identification

2. Feature Engineering

Created derived features to capture domain-specific insights.

  • • Minimum payment calculation
  • • Recency metrics
  • • Categorical binning
  • • Geographic encoding

3. Model Development

Systematic comparison and optimization of multiple algorithms.

  • • Cross-validation framework
  • • Hyperparameter optimization
  • • Performance evaluation
  • • Model interpretation

Website Development

This portfolio website was built to showcase the analysis in an interactive, accessible format. The development process focused on creating a professional presentation suitable for recruiters, collaborators, and technical audiences.

Design Principles

  • • Clean, professional aesthetic
  • • Responsive design for all devices
  • • Interactive visualizations
  • • Accessible navigation structure
  • • SEO optimization
  • • Performance optimization

Technical Implementation

  • • Next.js 15 with App Router
  • • TypeScript for type safety
  • • TailwindCSS for styling
  • • Recharts for data visualization
  • • Vercel for deployment
  • • Modern web standards

Built with Next.js, TypeScript, and TailwindCSS. Deployed on Vercel.

© 2024 Benjamin Gerochi. This project is for educational and portfolio purposes.