<![CDATA[This study develops an automatic classification system for customer complaints in the banking sector using the Long Short-Term Memory (LSTM) deep learning method. A dataset comprising 4,714 customer complaint entries was collected from Bank Sumut's internal communication records, categorized into six major complaint types. The data underwent comprehensive preprocessing, including cleaning, tokenization, and vectorization. A supervised learning approach was applied using an LSTM-based neural network architecture, and the model’s performance was evaluated using accuracy, precision, recall, and F1-score metrics. The results demonstrated a classification accuracy of 100% on the test set, with the model successfully categorizing free-text complaints into predefined categories. The findings highlight the strong potential of LSTM models in supporting automated text-based customer service operations within digital banking environments, particularly for Indonesian-language complaint datasets. Further research is recommended to validate the model on unseen real-world data and to address challenges related to data imbalance.]]>
Copyrights © 2025
