| Popis: |
Early prediction of heart disease is very important as diseases related to heart can turn out to be life-threatening. In this paper, a hybrid framework using unsupervised clustering technique with dimensionality reduction technique and regression technique is developed to predict the likelihood of presence of heart disease. Experimental results showed that our framework using k-means clustering, Principal Component Analysis (PCA) and Logistic Regression (LR) technique performed better, and 98.82% of accuracy has been achieved by the framework. The results are validated using tenfold cross validation. |
