Analisis Sentimen Pengunjung Terhadap Fasilitas Dan Layanan Di Pantai Pudak Menggunakan Algoritma Lstm

Abstract

This study aims to analyze visitor sentiment toward the facilities and services at Pudak Beach using the Long Short-Term Memory (LSTM) algorithm. The research adopts an applied study approach with text mining and deep learning-based sentiment analysis techniques. A total of 628 reviews were collected from Google Maps and processed through text preprocessing stages, including cleaning, case folding, tokenizing, stopword removal, and stemming, resulting in 419 clean data ready for analysis. The reviews were then classified into three sentiment categories: positive, neutral, and negative. The classification results showed that 306 reviews (73.03%) were positive, 69 reviews (16.47%) were neutral, and 44 reviews (10.5%) were negative. Model performance was evaluated using a confusion matrix with accuracy, precision, recall, and F1-score metrics. The test results showed that the LSTM model achieved an accuracy of 96% on the training data and 74% on the test data. The highest precision, recall, and F1-score were obtained in the positive sentiment category, while the neutral and negative categories performed lower due to imbalanced data distribution. This study provides a real picture of visitor perceptions and demonstrates the potential of applying big data and Natural Language Processing (NLP) in strategic decision-making in the tourism sector.