SmartMeals: Real-Time Institutional Food Demand Forecasting Using Hybrid LSTM with Attention and Temporal Feature Engineering

DOI

10.2991/978-94-6463-978-0_43How to use a DOI?

Keywords

Food demand forecasting; institutional kitchens; deep learning; LSTM attention mechanism time series prediction feature engineering·; Optuna real-time deployment smart supply chain

Abstract

Efficient prediction of institutional food demand is an essen- tial tool in enhancing the reduction of waste, improving procurement and sustainability for production kitchens including university campus dining services, military bases and public welfare organizations. Such day level forecasting cannot be carried out through traditional forecast- ing methods as those do not take into account the temporal, contextual and exogenous factors that strongly influence daily food eaten patterns. In this paper, we propose SmartMeals, a hybrid deep learning approach for short-term food demand forecasting employing an LSTM architec- ture enriched with convolutional layers, attention mechanisms as well as autotuning of hyperparameters powered by Optuna. Real-world aspects, e.g., weather conditions, holiday signs, day-of-week effects and lagged de- mand values are incorporated in the model to capture stable and volatile consumption behaviors. We train and evaluate the model on a simulated yet contextually realistic dataset and show that our MAPE for non- vegetarian demand is as low as 1.03%, while it’s even lower for vegetar- ian demand at 2.9% — substantially outperforming classical baselines. The complete system is implemented as a Dockerized micro-service with a REST API and front-end dashboard, enabling secure, hash-based real- time prediction traceability to any forecast request. The present work is methodologically practical and provides a scalable and interpretable direction for sustainable food supply planning in institutional settings.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Shivansh Gautam
AU  - Obillaneni Karthik Sree
AU  - R. Sapna
PY  - 2025
DA  - 2025/12/31
TI  - SmartMeals: Real-Time Institutional Food Demand Forecasting Using Hybrid LSTM with Attention and Temporal Feature Engineering
BT  - Proceedings of the 1st Engineering Data Analytics and Management Conference (EAMCON 2025)
PB  - Atlantis Press
SP  - 500
EP  - 512
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-978-0_43
DO  - 10.2991/978-94-6463-978-0_43
ID  - Gautam2025
ER  -