PERANCANGAN MODEL ANALISIS SENTIMEN LAYANAN LOKET.COM DI TWITTER DENGAN METODE TEXT MINING

• Main Author: Hanendya Ega Cristi K, Nicolaus
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77726
• Institution: Institut Teknologi Bandung

Loket.com is an e-commerce company providing online ticket sales services that experienced growth in 2023 due to the resumption of events post the Covid-19 pandemic. To facilitate continuous improvement in their business processes, Loket.com requires information about customer satisfaction levels and issues faced by customers. However, Loket.com lacks a direct application for gathering user feedback, necessitating the manual review of input. Leveraging the potential of the Twitter platform, which contains a significant volume of customer opinions and comments, a sentiment analysis model was designed to automatically process textual data and generate valuable insights. The design of the sentiment analysis model employed text mining techniques and followed the CRISP-DM methodology. The data collection process was conducted using tweet-harvesting, resulting in the acquisition of 8.296 text data from Twitter related to Loket.com. Subsequently, the data was cleaned and trained using machine learning algorithms such as CNN, LSTM, CNN-LSTM, LSTM-CNN, GRU- CNN, and SVM. Model evaluation demonstrated that the SVM algorithm, with an accuracy of 87% and a processing time of 5 seconds, outperformed other models, making it the chosen prototype. After being analyzed, it was concluded that a model with a simple architecture could predict moderately sized data more optimally. The prototype was developed using two platforms, namely Google Colab for data scraping and Streamlit for visualizing the prediction results of the sentiment analysis model. The application comprises three pages: the main page, data scraping page, and sentiment analysis page, all equipped with exploratory data analysis features. With a predictive sentiment analysis model in place, the company can easily gauge the sentiments of the Twitter community, serving as decision support to measure customer satisfaction levels and contribute to continuous improvement efforts in enhancing Loket.com's services.