Classification of Typhus and Dengue Fever Using the Pseudo Nearest Neighbor Algorithm

Abstract

Typhus and dengue fever are diseases that often occur in Indonesia. The spread of these two diseases is relatively fast with similar symptoms. This could be a fatal thing if there is a misdiagnosis. Therefore, an application was developed to assist in the classification of typhus and dengue fever based on the patient's clinical symptoms using the PNN (Pseudo Nearest Neighbor) algorithm. This application receives input in the form of clinical symptoms experienced by the patient, then a preprocessing process is carried out to convert user input into discrete data, and the results are processed in classification using the PNN method. From the validation process with 5-fold cross validation obtained the best k value is k=6. Then, the accuracy testing process concluded that the accuracy of the classification process for typhoid and dengue fever with the PNN method is 68,97%. Then, from 25 respondents in the user acceptance test obtained that 88.4% of respondents strongly agree with the application design, 87.6% respondents strongly agree with the ease of application, and 86.6% respondents strongly agree with the efficiency provided by the application.