Wine Analysis — SQL Mini Project

Team Slytherin — Eppu • Levent • María • Nodir • Lucie

---

🧠 Project Overview

We built a normalized MySQL database from a, previously cleaned by us, wine dataset and used SQL + Python to investigate how origin, closure type, and tasting characteristics relate to price. The workflow covers data loading (staging), transformation, modeling (ERD), and reproducible analysis.

---

🎯 Objectives (Hypotheses)

1. Where do the most expensive wines come from?
2. Does the closure (cork, screwcap, etc.) correlate with price?
3. What characteristics are shared by expensive wines?

---

🗺️ Data & Modeling

• Data prep: cleaned CSV → staging table → transformed into a normalized schema.
• Schema (core tables): main (facts), wine_info (titles, descriptions, vintage), lookups type, closure, primary_grape, and the many-to-many pair characteristic ↔ wine_characteristic.
• ERD approach: primary/foreign keys with wine_id at the center, enabling joins for region, closure, grapes, and characteristics.

---

🔍 Methodology

• Load cleaned CSV to staging (pandas / Workbench Wizard).
• Transform and normalize into the ERD tables.
• Write analytical SQL (GROUP BY, JOIN, WINDOW functions like NTILE) for premium tiers and comparisons.
• Validate with quick sanity checks (row counts, NULLs, types), then export compact result tables for presentation.

---

📊 Key Findings (Short)

• Premium origins: a consistent set of regions leads average prices when a minimum sample per region is enforced.
• Closures & price: natural cork wines are meaningfully pricier on average than screwcap/others.
• Expensive-wine traits: notes like citrus/green apple, vanilla/spice, bread/biscuit appear frequently among higher-priced bottles.
• Regional flavor signatures: cool areas skew citrus/green-apple; warmer red regions show black fruit + vanilla/spice; rosé regions show red fruit/strawberry/peach.
• Price concentration: a small top slice of wines drives most of total price (Pareto-style concentration).

---

🧰 Tech Stack

• Database: MySQL (Workbench for ERD/inspection; SQLAlchemy + PyMySQL from Jupyter).
• Python: pandas for cleaning and exports; Jupyter for the ETL + analysis notebook.

---

🗂️ Repository Structure

wine dataset/
├─ SQL queries compillation/
│
├─ canva slides/
│  
├─ data/
│
├─ erd/
│
├─ notebooks/

---


## 🙌 Credits
Team Slytherin — analysis, modeling, and presentation.

---