Scholarlink

Analyzing written customer reviews is something every business is focusing on these days. Written customer reviews are a valuable source of information that can provide insights into the system, but dealing with text feedback as unstructured data is difficult. This study utilizes advanced machine learning algorithms and statistics to enhance data quality and analyze customer feedback on Giao Hang Tiet Kiem services. Using a dataset of 12,624 customer evaluations, we applied particular methods such Support Vector Machine (SVM), Logistic Regression, Random Forest, Naive Bayes, and Decision Tree. The Decision Tree algorithm demonstrated the best accuracy at 0.97336 in the first results. Next, by utilising hyperparameter tuning, we were able to enhance the predictive performance and raise accuracy to 0.98205. Afterward, we created a web application that uses Streamlit technology to provide real-time sentiment analysis. This comprehensive approach provides useful insights into improving service quality and customer satisfaction.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Customer Sentiment Analysis: A Hybrid Approach Using VADER and Machine Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Truong Cong Doan,
  • Phan Thanh Duc,
  • Tran Hoang Son

摘要

Analyzing written customer reviews is something every business is focusing on these days. Written customer reviews are a valuable source of information that can provide insights into the system, but dealing with text feedback as unstructured data is difficult. This study utilizes advanced machine learning algorithms and statistics to enhance data quality and analyze customer feedback on Giao Hang Tiet Kiem services. Using a dataset of 12,624 customer evaluations, we applied particular methods such Support Vector Machine (SVM), Logistic Regression, Random Forest, Naive Bayes, and Decision Tree. The Decision Tree algorithm demonstrated the best accuracy at 0.97336 in the first results. Next, by utilising hyperparameter tuning, we were able to enhance the predictive performance and raise accuracy to 0.98205. Afterward, we created a web application that uses Streamlit technology to provide real-time sentiment analysis. This comprehensive approach provides useful insights into improving service quality and customer satisfaction.