Analysis of Visitor Sentiment towards Tourist Attractions in Gresik Regency Using Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA)

Abstract

 The tourism sector on the island of Java dominates the flow of domestic travel in Indonesia. East Java contributed the highest number with 198.91 million trips. However, this condition is still not evenly distributed across all regions. Based on the Online Tourist Visit Data (DAKUWISON), it was noted that there was a decrease in tourist visitors in Gresik Regency in 2023. This is not following the PPKM policy that was eliminated in the previous year. This research purposes to analyze the sentiment of reviews using the SVM-LDA classification method to determine their perceptions as an additional data-based opinion for tourism managers. Support Vector Machine (SVM) as a Supervised Learning method is applied in research, besides improving classification by adding the Linear Discriminant Analysis (LDA) dimension reduction method. Data collection from Google Maps with a web scraping technique obtained 3460 reviews. The results of research from the evaluation comparison of each model show that the SVM model with LDA is better than the SVM model without LDA. The f1-score value of the SVM model with LDA is 66% higher than the SVM model without LDA, with an f1-score value of 53%. Based on the results of sentiment classification on 2023 data, it shows that visitor sentiment tends to be positive, with 511 reviews, 456 positive sentiments, 33 negative sentiments, and 22 neutral sentiments obtained.