Wine Quality Analysis 🍷

Welcome to the Wine Quality Analysis project! In this repository, we dive into the world of unsupervised learning to uncover patterns and clusters in wine quality data. Whether you're a data enthusiast, a wine lover, or a machine learning practitioner, there's something here for you! 🍇🍷

---

📌 Project Overview

This project applies unsupervised learning techniques to analyze and cluster wines based on their chemical and physical properties. Using the popular Red Wine Quality Dataset, we aim to:

• Identify hidden patterns in the data.
• Group wines with similar characteristics.
• Gain insights into the features that influence wine quality.

---

🛠️ Methodology

Our approach is divided into four key steps, each documented in a dedicated Jupyter notebook:

1. Data Cleaning & Preprocessing:

   • Handle missing values and outliers.
   • Scale and normalize features for optimal clustering performance.

2. Exploratory Data Analysis (EDA):

   • Visualize distributions and correlations.
   • Understand feature importance using ANOVA and t-tests.

3. Unsupervised Learning:

   • Apply clustering algorithms like K-Means and DBSCAN.
   • Optimize the number of clusters using Elbow Method and Silhouette Score.

4. Dimensionality Reduction:

   • Use PCA to reduce feature dimensions and visualize clusters in 2D.

---

💻 Repository Structure

wine-quality-analysis/
├── data/                   # Dataset files
├── notebooks/              # Jupyter notebooks for each stage
│   ├── 1_data_cleaning.ipynb
│   ├── 2_exploratory_analysis.ipynb
│   ├── 3_statistical_analysis.ipynb
│   ├── 4_supervised_learning.ipynb
│   ├── 5_unsupervised_learning.ipynb
├── README.md               # Project overview
├── requirements.txt        # Python dependencies

---

🔬 Results

Clustering Performance:

• K-Means: Optimal number of clusters = 2, Silhouette Score = 0.20
• DBSCAN: Best parameters (eps=1.7, min_samples=8), Silhouette Score = 0.35

PCA Visualization:

Clusters are well-separated in 2D using the first two principal components, providing clear insights into the structure of the data.

---

🚀 Future Steps

1. Explore supervised learning approaches for predicting wine quality.
2. Develop an interactive Streamlit dashboard for data exploration.
3. Investigate other clustering algorithms like Agglomerative Clustering.

---

📚 References

This project was built using the following references and resources:

• Wine Quality Dataset - UCI Machine Learning Repository.
• "Wine Quality: Machine Learning in Practice" by Cortez et al., 2009.

---

📬 Contact

If you have any questions, suggestions, or feedback about this project, feel free to reach out!

• Author: Miguel Ángel Chamizo Sánchez
• Email: miguelchamizo10@gmail.com
• LinkedIn: https://www.linkedin.com/in/miguelangelchamizosanchez

Feel free to open an issue in this repository for any bugs or feature requests. Contributions are always welcome!

---

📦 Requirements

To set up the project environment, install the required dependencies:

pip install -r requirements.txt