MIS772 Predictive Analytics (2019 T1) Wine Rating Analysis Report

This report details an analysis of wine ratings conducted using RapidMiner for the MIS772 Predictive Analytics course. The study, commissioned by Australian Wine Importers (AWI), aims to predict wine ratings for new market entries. The analysis involves data preparation, including handling missing values and discretizing rating points. Three predictive models were built: decision tree, k-NN, and gradient boosting tree, with k-NN emerging as the best performer. The report includes model evaluation using metrics like accuracy and Kappa, followed by model improvement through cross-validation and parameter tuning. An integrated solution was developed to predict the rating of new wines, with k-NN model demonstrating maximum effectiveness. Further research extended the analysis by implementing a k-NN model in R. The report concludes with a bibliography of relevant data mining and analytics resources.