Performance Improvement of The Random Forest Method Based on Smote-Tomek Link on Lombok Tourism Analysis Sentiment

Abstract

Background: Tourists visiting Lombok Island can access various sources of tourist information and can share their views and tourist experiences through social media such as positive and negative experiences.Objective: This research aims to analyze the sentiment of Lombok tourism reviews using the Smote-Tomek Link and Random Forest algorithms.Methods: The research was carried out in several stages, namely collecting the Lombok tourism dataset, text preprocessing, text weighting using the Term Frequency-Inverse Document Frequency (TF-IDF) method, data sampling using SMOTE-Tomek Link, text classification using Random Forest, and the final stage was performance testing based on accuracy.Result: The research results obtained using the Smote-Tomek Link and Random Forest methods in sentiment analysis analysis of tourist reviews about Lombok were 94%.Conclusion: The use of the Smote-Tomek Link and Random Forest methods in Lombok tourism sentiment analysis produces very good accuracy.