Comparative analysis of decision tree and support vector machine algorithm in sentiment classification for birds of paradise content

Abstract

This research aims to analyze public sentiments towards National Geographic's content on the bird of paradise from the perspective of nature-based tourism. The method utilized is CRISP-DM, comprising stages of business understanding, data understanding, modeling, evaluation, and deployment. Focusing on sentiments expressed in response to National Geographic's Bird of Paradise content, this study seeks insights into how the public perceives and values nature-oriented tourism experiences. Comparing the results of DT and SVM algorithms with and without the SMOTE reveals noteworthy differences in classification performance. Without SMOTE, both DT and SVM exhibit relatively lower accuracy and AUC values compared to their counterparts with SMOTE. For DT, adding SMOTE substantially improves accuracy (from 92.44% to 95.20%) and AUC (from 0.517 to 0.956), indicating enhanced classification accuracy and model robustness. In addition, SVM demonstrates significant performance gains with SMOTE, achieving notably higher accuracy (from 92.12% to 98.63%) and AUC (from 0.617 to 0.999). The significantly higher values across various performance metrics for SVM underscore its effectiveness in handling imbalanced datasets and accurately classifying sentiment data. Therefore, researchers and practitioners may consider leveraging SVM for sentiment analysis tasks in similar contexts to achieve optimal classification results and enhance decision-making processes.