An end-to-end data analytics pipeline analyzing 251,000+ Airbnb listings across 10 European cities. From automated web scraping with Playwright to real-time price prediction with CatBoost, this project covers every stage of the data lifecycle: collection, cleaning, warehousing, statistical testing, visualization, and deployment.
Built with Python, R, SQL Server, Power BI, and Streamlit. The live web app features an interactive map, 7 real-time filters, 6 analytical charts, and an AI-powered price predictor that estimates nightly rates based on city, room type, guest capacity, cleanliness, distance to center, and day type.
The pipeline starts with two custom Playwright scrapers (CLI and GUI) that collect listing data from Airbnb with anti-bot stealth mode. Raw data is then cleaned using Pandas with KNN imputation for missing values, normalized, and feature-engineered into 30+ columns. The clean dataset is loaded into a SQL Server star schema with 5 dimension tables and 1 fact table, enabling 9 advanced analytical queries covering listing scoring, deal detection, amenity tier classification, and geographic demand hotspots.
Statistical analysis in R confirms significant price differences across room types using ANOVA with Tukey HSD post-hoc testing, while MANOVA reveals multivariate effects on price, satisfaction, and cleanliness simultaneously. A 4-page Power BI dashboard provides interactive business intelligence visualizations. The Streamlit web app, deployed live, combines all insights with a CatBoost regression model that predicts nightly prices from 6 key features, allowing users to click any listing on the map and get an instant price estimate.
