Prediction of Hypertension in the Upcoming Year: Feature Correlation Analysis and Handling Imbalanced Based on Random Forest

Hypertension is a disease caused by a person’s high blood pressure. If left unchecked, this condition will lead to other serious complications such as heart disease, stroke, and even death. It is important to take preventive measures to reduce a person’s risk of developing hypertension. One of the steps that can be taken is to predict a person’s risk of hypertension in the coming year using a collection of clinical datasets (general checkup). This study has the main objective of predicting a person’s risk of developing hypertension in the coming year based on the general checkup (GCU) dataset. The GCU dataset in this study is the result of a clinical examination of a person every year for three years (2019-2021). Analysis between features in the dataset was also carried out to find out which features have a high correlation with hypertension. Apart from that, other preprocessing treatments were carried out, namely label encoding, missing values and imbalanced. The handling of missing values uses the statistical mean technique, while the handling of imbalanced uses the Random Oversampling technique. Several machine learning models such as Decision Tree, Random Forest, Adaptive Boosting (AdaBoost), Xtreme Gradient Boosting (XGBoost), Regression Logistics, Naïve Bayes, Support Vector Machine, and Multi-Layer Perceptron are applied to predict hypertension. The Random Forest model produces the best performance with an accuracy of 97.5% and a precision of 73.9%. These findings help prevent hypertension and change to a healthier lifestyle.