| Summary: | Understanding customer food preferences is vital for restaurants to offer valuable services. Recognizing these preferences and the underlying sentiments helps in understanding how customers make dining decisions, where reviews on food type, pricing, portion size, and taste play a crucial role. This study focuses on local, Western, and Asian cuisines, emphasizing the importance for restaurant owners to stay informed about consumer demands and how sentiments influence food choices. It aims to identify sentiments and keywords associated with preferred food types from customer tweets and to determine whether positive sentiments have a greater impact than negative ones in the decision-making process. The research uses a machine learning approach, incorporating bag-of-grams models, sentiment analysis, and hypothesis testing for proportions, with data collected from Twitter between April and May 2023. Results show that the Naive Bayes model had the highest accuracy (72.50%) compared to k-NN (67.8%) and Decision Tree (61.85%). Sentiment analysis also indicates that positive sentiments have a stronger influence on customer decisions than negative ones. © 2024 IEEE.
|
|---|