Pembobotan Kriteria Dalam Prediksi Meningitis Tuberkulosis Menggunakan Metode SWARA dan Nearest Neighbor

Abstract

Weights greatly affect the value and results of decisions or predictions of a test data, a problem that often occurs in the results of the prediction process is the weighting of symptom attributes which is less certain of the value of the weight, thus affecting the prediction results and the level of accuracy of a prediction itself. This study predicts a data using the Nearest Neighbor method where in the process of predicting the attribute weight value does not yet have a definite value for testing. Then we need an attribute weighting for each test attribute to get a definite weight value result. One method that can be applied to attribute weighting is the SWARA method. Based on research conducted to compare the prediction of Meningitis Tuberculosis without SWARA weighting and with SWARA weighting, testing with a ratio of 90:10, 80:20, 70:30 results in disease prediction using the Nearest Neighbor method, there are differences in results and levels of prediction accuracy and the process in prediction helps shorten the time to find prediction results, the highest prediction result using the swara method is 100% accurate and without weighting method is 91%.