Comparison of k-Nearest Neighbor and Support Vector Machine using Binary Dragonfly Algorithm Optimization

BDA is an adaptation of Dragonfly Algorithm (DA) that optimizes computation for single-objective, discrete, and multi-objective problems. Combining BDA optimization algorithm with KNN and SVM classification algorithms aims to improve the performance of the prediction model. This research compares and tests accuracy of KNN and SVM algorithms on the diabetes dataset used in research to find out the best algorithm in predicting diabetes. This research uses the BDA optimization algorithm to select the best features in the dataset, then the KNN and SVM classification algorithms, in classifying data, predicting, and comparing the accuracy of the accuracy of the two algorithms on the diabetes dataset. Medical record data from people with diabetes is processed using the KNN and SVM algorithms, which will then produce an accuracy level that can be used in predicting diabetes. Previous research has conducted a comparison between classification algorithms in predicting diabetes. In the previous research above, no one has combined BDA with classification algorithms, because BDA itself is a relatively new method and has not been widely studied, so researchers use this optimization algorithm. The results of the research conducted obtained the highest accuracy results in the BDA + KNN algorithm with a Precision value of 96.10%, Recall 79.36%, F-1 Score 86.93% and Accuracy 85.55%.