Wine-Classifier

This project is a multiclass classification model designed to predict the quality of wine based on physicochemical properties. Using the Wine Quality Dataset (UCI), the model classifies wine samples into quality ratings ranging from 3 to 8, providing a practical demonstration of machine learning in quality control for food and beverage industries.

Model: Multiclass Classification (Random Forest)

Dataset: Wine Quality Dataset (UCI)

Preprocessing:

Handled class imbalance (e.g., class merging or resampling techniques)

Normalized features to improve convergence

Performed EDA to identify key feature distributions

Evaluation:

Used a confusion matrix and classification report for insight into per-class performance

Random Forest Model

Multiclass AUC: 0.9892

          precision    recall  f1-score   support

       0       0.99      0.99      0.99       291
       1       0.92      0.98      0.95       272
       2       0.82      0.78      0.80       287
       3       0.73      0.67      0.70       268
       4       0.86      0.89      0.88       281
       5       0.95      0.97      0.96       312
       6       1.00      1.00      1.00       272

accuracy                           0.90      1983

macro avg 0.90 0.90 0.90 1983 weighted avg 0.90 0.90 0.90 1983

Objective:

To develop a robust model capable of automatically grading wine quality based on measurable chemical attributes, demonstrating real-world applications of machine learning in classification tasks.