Performance Comparison of Support Vector Machine and Deep Neural Network for Sentiment Classification in Digital Tourism

This study aims to classify public sentiment on SNS Instagram @indtravel content using Machine Learning techniques with Support Vector Machine (SVM) and Deep Neural Network (DNN) Modeling. From the results of this analysis, it can be seen whether the tourism promotion strategy by the Indonesian Ministry of Tourism through SNS Instagram tends to be positive or negative towards the push and pull factors for someone to take a tour. In addition, this study also aims to compare the performance ​​of Accuracy, Precision, Recall, F1-Measure, ErrorRate, and the AUC-ROC Curve of the SVM and DNN models. The dataset used in this study was obtained from SNS Instagram @indtravel comments using scraping techniques. The results of the evaluation in this study indicate that the public sentiment towards the content of SNS Instagram @indtravel tends to be positive towards the push and pull factors for someone to take a tour. Based on the results of the performance comparison between the SVM and DNN model, it is proven DNN model has a higher level of performance in Accuracy (89,37%), Precision (93,79%), F1-Measure (93,79%), ErrorRate (10,63%), were the SVM model only higher in Precision rate with a difference of 5,43%. This indicates that the DNN model has a very good performance in classifying public sentiment on SNS media compared to the SVM model