<![CDATA[High levels of sugar in the blood can cause diabetes. The longer people are unable to control glucose in their blood, the more complications it can cause, other diseases and even death. Early detection of diabetes is needed, one way is by carrying out data mining classification. Data mining classification in this research uses two algorithms, namely SVM (Support Vector Machine) and Naïve Bayes. This research compares the two algorithms using two methods, namely training split and k-fold cross validation which aims to get the best classification results in detecting diabetes. The best classification results are determined by calculating the average value of precision, recall and accuracy. Based on this research, the SVM algorithm with split percentage training produces average values for precision, recall and accuracy, namely 77%, 71.5%, 77.27%, while the SVM algorithm with k-fold cross validation produces average values for precision, recall , and accuracy is 77%, 72.5%, 71%. The Naïve Bayes algorithm with the split percentage training method produces average values for precision, recall and accuracy, namely 75.5%, 74.5%, 79%, while the Naïve Bayes algorithm with k-fold cross validation produces average values for precision, recall, and accuracy of 75.5%, 74.5%, 75%. The best classification result in detecting diabetes is the Naïve Bayes algorithm, the split percentage method, which provides the best accuracy, precision and recall values above 74%.]]>
Copyrights © 2024
