HEART DISEASE PREDICTION SYSTEM USING ENSEMBLE PREDICTIVE MODELLING WITH PRINCIPAL COMPONENT ANALYSIS

In the recent years, data mining has been employed in the medical field for extracting and manipulating information, and aids within the higher process. Predicting the results of a process with a high level of accuracy is a difficult task. In this work, the advantage of the data mining models have been taken to predict the heart disease. The benchmark dataset, "Cleveland Heart Disease" dataset from UCI machine learning repository has been used. The main objective of this work is to propose an extensive data pre-processing task such as imputation of missing values and a feature engineering technique namely,Principal Component Analysis. to be used to transform the dataset in a compressed form. Ensemble / classifier combination method called boosting method such as Gradient boosting machine and Random Forest are used. The results show that, the ensemble learners, with PCA attained ROC - AUC value of 0.90 and above with 100% accuracy of prediction. Moreover, it ensures that no missing information must be removed and might be imputed to confirm the data quality is enough. As a result, the model has shown to be helpful for the real time prediction of heart disease.