Year: 2025
Type: article
Abstract: Nowadays, it's fairly easy to browse menus, place orders, and use meal delivery applications like Zomato and Swiggy, your favorite meals, and leave ratings, food from different restaurants. These ratings and reviews are helpful not just for customers but also for businesses. However, figuring out the overall sentiment from these reviews can be tricky. To better understand the data, we did some exploratory analysis to identify the most and least expensive restaurants. We also found the top critics—those with more than 100 reviews and 10,000 followers. We then used clustering methods like KMeans and Hierarchical clustering to be grouped restaurants into three categories based on their cuisine type and pricing. For sentiment analysis, we tried both supervised methods (like Logistic Regression, Decision Trees, and Naive Bayes) and unsupervised methods (like Linear Discriminant Analysis). We defined ratings above 3.5 as positive. After some fine-tuning, we found that Logistic Regression and LightGBM worked the best.
Authors: Swathi Turai, Praneetha. P, Rajasri Aishwarya. B, Mohammed Adil, Mani Charan Vangala
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4407590016