Performance Evaluation of Quantitative Data Analysis Methods in the Prediction of Angiographic Disease Status

摘要

Data Analytics play a significant role in arriving at crucial conclusions by sifting large volumes of data. Predictive analytics in healthcare is taking a new paradigm shift with the advent of efficient data mining techniques to identify interesting patterns which otherwise go unnoticed. Ensemble methods boost this process by forming strong classification tree models which perform better than its component models. Mining health datasets require high degree of precision to arrive at valid conclusions and selecting the best classifier is a critical task for that. This present work evaluates the classification models namely Decision Tree, Naïve Bayes, k-Nearest Neighbour, Rule Induction and Random Forest to find the best performing models to predict the Angiographic Disease status that is important in diagnosing heart disease. Considering the ROC curve, optimal performance is shown by Naïve Bayes and Random Forest among the selected classifiers.