A Highly Optimum, Efficient and Accurate Model to Predict Wine Quality using Multi-Class Classification

The purpose of this report is to predict the quality of white wine using machine learning algorithms. In addition to this, we demonstrate our ability to explore and process the data, and to implement multiple machine learning models on given datasets to achieve optimum results. In this report, the method of multi-class classification is employed to distinguish classes from each other. A labeled dataset of 4898 instances was used to predict the quality of the wine based on provided 11 features. Multiple models were employed and tested on the labeled dataset from simple models, like K Nearest-Neighbors to more complex models such as Random Forest and Neural Networks. The performance of these models were compared based on their performance on a validation dataset. The best accuracy of 54% was obtained using the Random Forest mode