MIS772 Predictive Analytics (2019 T1) Assignment A2: Wine Prediction

This project, completed for MIS772 Predictive Analytics, focuses on predicting wine ratings using a dataset of 130,000 wine test results. The student employed RapidMiner to preprocess the data, including handling missing values and categorizing rating points. Three predictive models—Decision Tree, Gradient Boost Tree, and k-NN clustering—were developed, trained, and evaluated using accuracy, classification error, Kappa, and R2 metrics. The k-NN model demonstrated the best performance and was further refined by optimizing the k value. The integrated solution predicted the rating of newly introduced wine, concluding it would likely be marked as excellent. The project also extended the research by implementing a k-NN model in R, achieving 95% accuracy, and includes a bibliography of relevant sources. The project provides a comprehensive analysis of wine data, showcasing predictive modeling techniques in data science.