Smart Dining: A Machine Learning Approach to Caloric Display in Bangladeshi Restaurants

Abstract

This research applies machine learning methods to estimate and present the calorie content of traditional Bangladeshi cuisine, empowering individuals to make informed dietary choices and encouraging healthier eating habits. The study involves a meticulously curated dataset of ingredients and recipes for training machine learning models. Three models, Linear Regression, Decision Tree, and Random Forest regressions are evaluated based on their performance using metrics such as Mean Squared Error (MSE) and R-squared (R²). Computational tools consolidate ingredient calorie data, integrate datasets, and compute total calories for each recipe. Data is split into training and testing sets, features are engineered, and models are trained and evaluated. Visualization metrics compare model accuracy, including scatter plots and bar charts. The analysis identifies Linear Regression as the top-performing model, achieving the lowest MSE of 15.75 and the highest R² score of 0.9999, indicating high predictive accuracy. The research adheres to ethical data practices and proposes a sustainability plan to mitigate the environmental impact of computational processes. By integrating machine learning into restaurant menus, this study offers an effective means of providing caloric information, thus contributing to global health goals focused on preventing diet-related health issues. Additionally, the research suggests opportunities for future studies to examine the societal acceptance and adoption of such technologies within Bangladeshi culture. This innovation has the potential to empower consumers to make healthier dietary decisions and foster a culture of healthy eating, thereby enhancing public health.