Analyzing Customer Satisfaction Through Face Emotion Recognition: A Comparative Study of Convolutional Neural Networks (CNN) and Long Short Term Memory (LSTM)
Dody Harianto, Stefan Filbert, Arya Baskara Cahyakusuma, Alfi Yusrotis Zakiyyah · 2024
Customer satisfaction is a key factor in all industry sectors. There are various techniques to analyze customer satisfaction, ranging from classic methods such as surveys to modern methods such as machine learning. To that end, this study aims to compare the best deep learning model in analyzing customer satisfaction by exploiting the capabilities of deep learning models. We propose and build a Convolutional Neural Network (CNN) and one of the main building blocks of transformers, Long Short-Term Memory (LSTM), to recognize and classify facial expression, which implementation can be used in detecting customer satisfaction levels. The process entails training the CNN and LSTM on a variety of datasets in order to achieve accurate and robust emotion recognition. Our results show that the proposed approach is effective, with CNN obtaining an excellent accuracy rate of 97.26% in accurately identifying customer emotions. Similarly, the LSTM model performed well, with an accuracy rate of 92.38%, suggesting its ability to capture the temporal dynamics of facial expressions.