Red wine classification

This study consider the Red-wine dataset of Portuguese "Vinho Verde" wine. The dataset contents 1599 samples of wine described by 11 physicochemical properties and a quality score. The task at hand is to predict the quality target variable from the 11 physicochemical features. We consider both regression and classification algorithm and compare results in order to provide an overall overview of Machine Learning performance on this dataset. Results are improved by a Grid-search strategy on grid parameter with a Cross-validation.

Outline:

• 1. Data insights
• 2. Data processing
• 3. Naive baseline with k-NN Regressor
• 4. Classification models
• 5. Conclusion