Perbandingan Nilai K pada Klasifikasi Pneumonia Anak Balita Menggunakan K-Nearest Neighbor

Abstract

Pneumonia is an infectious disease that attacks the lower respiratory tract and is one of the main causes of death in children under five. Pneumonia is easy to attack toddlers caused by various microorganisms that exist in the environment such as viruses, bacteria, fungi and micro bacteria. This study uses K-Nearest Neighbor (KNN) for the classification of pneumonia in patients based on the symptoms experienced. The KNN classification method is carried out by comparing the object distance between the test data and the overall object on the training data based on the patient's medical history data. The comparison of the percentage of the training data and the test data used is 90:10, 80:20, and 70:30 to calculate the value of the closest distance of the test data to the overall training data with the number of k used. The confusion matrix was used to measure the results of the Pneumonia classification test for toddlers with a combination of the amount of training data and test data on the number of k={1, 3, 5, 7, 9, 11}, the highest accuracy, precision, recall, and F-measure values were obtained. 0.86, 0.89, 1, and 0.91 for 90% training data, 10% test data with a value of k = 3. Keywords: Pneumonia, Toddlers, KNN, Confusion Matrix.