Optimization of the Naïve Bayes Algorithm Using Ant Colony Optimization for the Classification of Stroke Patient Data

Stroke is a dangerous disease that is included in the 10 deadliest diseases in Indonesia, with a very high mortality rate. To detect strokes, CT scans and MRI examinations are usually used. However, this examination is expensive and is often only carried out after someone has had a stroke. As a result, medical treatment can be delayed and the patient's condition can worsen. Therefore, it is very important for a method that can identify a patient's condition at an early stage to detect the risk of stroke by classifying whether the patient is experiencing a stroke or not. One method that is often used for classification is Naive Bayes. However, this method has the weakness of passing data into certain classes even though the data is not relevant or relevant so it needs to be optimized by feature selection. From this, optimization was carried out using the Ant Colony Optimization algorithm for feature selection for classification using Naive Bayes. In this research, feature selection will be carried out on the data that will be classified with Naive Bayes using the Ant Colony Optimization algorithm. The accuracy of the Naive Bayes method before and after being optimized with the Ant Colony Optimization algorithm was 86.45% and 95.84% without the use of resampling techniques, while with the application of resampling techniques, the accuracy was 78.69% and 79.91 %.