Implementasi Particle Swarm Optimization (PSO) Pada Analysis Sentiment Review Aplikasi Halodoc Menggunakan Algoritma Naïve Bayes

Abstract

Health is very important for humans, if you experience symptoms or feel pain, it is appropriate for us to have a health check and go to a hospital or clinic, but if it is not possible to leave the house, an online health consultation application is considered to be helpful. But before you can use and take advantage of these applications, it is necessary to know reviews from consumers based on positive opinions and negative opinions. This study applies the Naive Bayes algorithm to perform text classification and selects the particle swarm optimazation selection feature to support the increased accuracy obtained. Classification evaluation and validation are performed using confusion matrix and ROC curves. The results showed an increase in accuracy previously 88.50% and AUC 0.535, increased to 90.50% and AUC 0.525. It can be concluded that the selection of the particle swarm optimazation feature has succeeded in increasing the accuracy.Keywords: selection features, naïve bayes, particle swarm optimization.