PENGARUH SMOTE TERHADAP PERFORMA ALGORITMA RANDOM FOREST DAN ALGORITMA GRADIENT BOOSTING DALAM MEMPREDIKSI PENYAKIT STROKE

Abstract

Stroke is an illness that can happen unexpectedly, swiftly leading to progressive brain damage due to non-traumatic disruptions in cerebral blood flow. a common symptom in stroke sufferers is numbness in the limbs, This condition ranks as the second leading cause of death globally and a significant contributor to cognitive impairment, placing third worldwide. Hence, recognizing the initial signs of a stroke is crucial for early prevention and intervention. Advancements in medical technology enable the application of machine learning for stroke prediction. Machine learning algorithms can deliver precise forecasts. This research utilizes two algorithms, namely Random Forest and Gradient Boosting, to predict stroke, with the goal of determining the most efficient algorithm for predicting strokes.The SMOTE oversampling method is applied to handle class imbalance in the dataset. The research results revealed that the Random Forest Algorithm model was superior in predicting stroke with an accuracy of 95.5%, precision of 78.8%, recall of 93.1%, and f1-score of 84.2%.