Wine Quality Classification

A machine learning project that predicts wine quality using physicochemical properties of Portuguese "Vinho Verde" red wine.

📊 Dataset

The dataset contains 1,599 samples of red wine with 11 physicochemical features and 1 quality rating:

Features:

1. Fixed acidity
2. Volatile acidity
3. Citric acid
4. Residual sugar
5. Chlorides
6. Free sulfur dioxide
7. Total sulfur dioxide
8. Density
9. pH
10. Sulphates
11. Alcohol

Target Variable:

• Quality: Score between 0-10 (converted to binary: Good ≥7, Bad <7)

🚀 Models Implemented

1. Random Forest Classifier

• Test Accuracy: 92.5%
• Cross-validation Score: 86.4% (±0.016)
• Optimized with GridSearchCV

2. Logistic Regression

• Test Accuracy: 89.4%
• Cross-validation Score: 87.0% (±0.019)
• Optimized with GridSearchCV

3. Decision Tree Classifier

• Test Accuracy: 90.3%
• Cross-validation Score: 79.8% (±0.060)

🔧 Key Features

• Data Preprocessing: Label binarization for quality classification
• Feature Selection: Top 5 most important features identified
• Hyperparameter Optimization: GridSearchCV for model tuning
• Cross-validation: 5-fold CV for robust evaluation
• Visualization: Correlation heatmaps and feature analysis

📈 Top 5 Most Important Features

1. Alcohol (17.7% importance)
2. Sulphates (11.7% importance)
3. Volatile Acidity (11.0% importance)
4. Citric Acid (9.7% importance)
5. Density (9.3% importance)

🛠️ Installation & Usage

Prerequisites

pip install numpy pandas matplotlib seaborn scikit-learn

Running the Code

# Clone the repository
git clone https://github.com/beater35/ml-classification-wine-quality.git
cd ml-classification-wine-quality

# Load and run the Jupyter notebook
jupyter notebook classification_model_comparison_wine_quality.ipynb

Making Predictions

# Example prediction with top 5 features
input_data = [10.0, 0.47, 0.65, 0.0, 0.9946]  # [alcohol, sulphates, volatile_acidity, citric_acid, density]
prediction = final_rf_model.predict([input_data])

if prediction[0] == 1:
    print("Good Quality Wine")
else:
    print("Bad Quality Wine")

📊 Model Performance Summary

Model	Test Accuracy	CV Score (Mean ± Std)
Random Forest	92.5%	86.4% ± 1.6%
Logistic Regression	89.4%	87.0% ± 1.9%
Decision Tree	90.3%	79.8% ± 6.0%

🔍 Project Structure

wine-quality-classification/
├── classification_model_comparison_wine_quality.ipynb
└── README.md

📝 Methodology

1. Data Exploration: Statistical analysis and visualization
2. Preprocessing: Binary classification setup (Good/Bad quality)
3. Model Training: Three different algorithms tested
4. Hyperparameter Tuning: GridSearchCV optimization
5. Feature Selection: Importance-based feature ranking
6. Model Evaluation: Cross-validation and test accuracy

🎯 Results

The Random Forest Classifier achieved the best performance with 92.5% test accuracy after hyperparameter optimization and feature selection. The model successfully identifies wine quality based on physicochemical properties, with alcohol content being the most influential factor.

📚 References

• Dataset: Kaggle – Wine Quality Dataset