Machine Learning Techniques for Cafeteria Demand Forecasting:  An Institutional Case

Büşra Aydın; Yavuz Selim Balcıoğlu; Bülent Sezen

Article Details

Machine Learning Techniques for Cafeteria Demand Forecasting: An Institutional Case

Authors:

Büşra Aydın | Yavuz Selim Balcıoğlu | Bülent Sezen

Publisher:

İdeal Kent Yayınları

Main Topics:

Yazılım Mühendisliği (Diğer)

Journal:

OPUS Toplum Araştırmaları Dergisi Vol.22 Issue 3, pp.532-549

Document Type:

Article

Publication Date:

5/2025

Document Language:

English

Document ID (Org No.):

2001348662

pISSN:

2791-9862

Keywords:

Abstract:

 This research investigates the application of machine learning techniques for cafeteria demand forecasting within institutional settings, addressing critical operational challenges in food service management. Using a comprehensive methodological framework, the study analyzes turnstile entry data from an academic institution across November-December 2023 to develop and evaluate three complementary forecasting models: XGBoost with time-based features, Long Short-Term Memory LSTM networks, and Prophet models with domain-specific components. The comparative analysis reveals differentiated performance characteristics across various forecasting dimensions, with XGBoost demonstrating superior accuracy for daily forecasting MAE16.23, MAPE8.32, LSTM excelling at high-resolution 15-minute interval prediction MAE5.37, MAPE11.64, and Prophet exhibiting greater stability for extended forecast horizons. A weighted ensemble methodology integrating these complementary approaches yields consistent performance improvements across multiple evaluation metrics, achieving 4.7 reduction in daily MAE and 3.4 reduction in 15-minute interval MAE compared to the best individual models. Feature importance analysis reveals the significance of recent historical patterns, weekly cyclical components, and academic calendar effects, validating the theoretical multi-level temporal structure of institutional demand. The operational impact assessment demonstrates substantial potential benefits, including estimated food waste reduction of 6.2, enhanced service level maintenance, and improved resource utilization. This research contributes methodological advancements through its multi-resolution forecasting framework, systematic feature engineering approach, and context-sensitive ensemble integration methodology, while providing practical implementation guidance for institutional food service operations seeking to enhance operational efficiency and sustainability through improved demand prediction.

 This research investigates the application of machine learning techniques for cafeteria demand forecasting within institutional settings, addressing critical operational challenges in food service management. Using a comprehensive methodological framework, the study analyzes turnstile entry data from an academic institution across November-December 2023 to develop and evaluate three complementary forecasting models: XGBoost with time-based features, Long Short-Term Memory LSTM networks, and Prophet models with domain-specific components. The comparative analysis reveals differentiated performance characteristics across various forecasting dimensions, with XGBoost demonstrating superior accuracy for daily forecasting MAE16.23, MAPE8.32, LSTM excelling at high-resolution 15-minute interval prediction MAE5.37, MAPE11.64, and Prophet exhibiting greater stability for extended forecast horizons. A weighted ensemble methodology integrating these complementary approaches yields consistent performance improvements across multiple evaluation metrics, achieving 4.7 reduction in daily MAE and 3.4 reduction in 15-minute interval MAE compared to the best individual models. Feature importance analysis reveals the significance of recent historical patterns, weekly cyclical components, and academic calendar effects, validating the theoretical multi-level temporal structure of institutional demand. The operational impact assessment demonstrates substantial potential benefits, including estimated food waste reduction of 6.2, enhanced service level maintenance, and improved resource utilization. This research contributes methodological advancements through its multi-resolution forecasting framework, systematic feature engineering approach, and context-sensitive ensemble integration methodology, while providing practical implementation guidance for institutional food service operations seeking to enhance operational efficiency and sustainability through improved demand prediction.

 This research investigates the application of machine learning techniques for cafeteria demand forecasting within institutional settings, addressing critical operational challenges in food service management. Using a comprehensive methodological framework, the study analyzes turnstile entry data from an academic institution across November-December 2023 to develop and evaluate three complementary forecasting models: XGBoost with time-based features, Long Short-Term Memory LSTM networks, and Prophet models with domain-specific components. The comparative analysis reveals differentiated performance characteristics across various forecasting dimensions, with XGBoost demonstrating superior accuracy for daily forecasting MAE16.23, MAPE8.32, LSTM excelling at high-resolution 15-minute interval prediction MAE5.37, MAPE11.64, and Prophet exhibiting greater stability for extended forecast horizons. A weighted ensemble methodology integrating these complementary approaches yields consistent performance improvements across multiple evaluation metrics, achieving 4.7 reduction in daily MAE and 3.4 reduction in 15-minute interval MAE compared to the best individual models. Feature importance analysis reveals the significance of recent historical patterns, weekly cyclical components, and academic calendar effects, validating the theoretical multi-level temporal structure of institutional demand. The operational impact assessment demonstrates substantial potential benefits, including estimated food waste reduction of 6.2, enhanced service level maintenance, and improved resource utilization. This research contributes methodological advancements through its multi-resolution forecasting framework, systematic feature engineering approach, and context-sensitive ensemble integration methodology, while providing practical implementation guidance for institutional food service operations seeking to enhance operational efficiency and sustainability through improved demand prediction.

 Bu araştırma, kurumsal ortamlarda kafeterya talep tahmini için makine öğrenimi tekniklerinin uygulanmasını araştırmakta ve gıda hizmeti yönetimindeki kritik operasyonel zorlukları ele almaktadır. Kapsamlı bir metodolojik çerçeve kullanan çalışma, üç tamamlayıcı tahmin modeli geliştirmek ve değerlendirmek için Kasım-Aralık 2023 tarihleri arasında bir akademik kurumdan alınan turnike giriş verilerini analiz etmektedir: Zaman tabanlı özelliklere sahip XGBoost, Uzun Kısa Süreli Bellek LSTM ağları ve alana özgü bileşenlere sahip Prophet modelleri kullanılmıştır. Karşılaştırmalı analiz, çeşitli tahmin boyutlarında farklılaşan performans özelliklerini ortaya koymaktadır; XGBoost günlük tahmin için üstün doğruluk gösterirken MAE  16.23, MAPE  8.32, LSTM yüksek çözünürlüklü 15 dakikalık aralık tahmininde mükemmeldir MAE  5.37, MAPE  11.64 ve Prophet genişletilmiş tahminler için daha fazla istikrar sergilemektedir. Bu tamamlayıcı yaklaşımları entegre eden ağırlıklı bir topluluk metodolojisi, birden fazla değerlendirme ölçütünde tutarlı performans iyileştirmeleri sağlamakta ve en iyi bireysel modellere kıyasla günlük MAEde 4,7 ve 15 dakikalık aralık MAEde 3,4 azalma göstermektedir. Özellik önem analizi, kurumsal talebin teorik çok seviyeli zamansal yapısını doğrulayarak yakın geçmişteki tarihsel modellerin, haftalık döngüsel bileşenlerin ve akademik takvim etkilerinin önemini ortaya koymaktadır. Operasyonel etki değerlendirmesi, tahmini 6,2lik gıda atığı azaltımı, gelişmiş hizmet seviyesi bakımı ve iyileştirilmiş kaynak kullanımı dahil olmak üzere önemli potansiyel faydalar göstermektedir. Bu araştırma, çok çözünürlüklü tahmin çerçevesi, sistematik özellik mühendisliği yaklaşımı ve bağlama duyarlı topluluk entegrasyon metodolojisi aracılığıyla metodolojik ilerlemelere katkıda bulunurken, gelişmiş talep tahmini yoluyla operasyonel verimliliği ve sürdürülebilirliği artırmak isteyen kurumsal gıda hizmeti operasyonları için pratik uygulama kılavuzu sağlamaktadır.

Article Details

Machine Learning Techniques for Cafeteria Demand Forecasting: An Institutional Case

Authors:

Büşra Aydın | Yavuz Selim Balcıoğlu | Bülent Sezen

Publisher:

İdeal Kent Yayınları

Main Topics:

Yazılım Mühendisliği (Diğer)

Journal:

OPUS Toplum Araştırmaları Dergisi Vol.22 Issue 3, pp.532-549

Document Type:

Article

Publication Date:

5/2025

Document Language:

English

Document ID (Org No.):

2001348662

pISSN:

2791-9862

Machine learning

Demand forecasting

Institutional catering

Time series analysis

Food waste reduction.Makine öğrenmesi

Talep tahmini

Kurumsal yemek hizmeti

Zaman serisi analizi

Yemek israfını azaltma

Issues

Article Details

Authors: Büşra Aydın | Yavuz Selim Balcıoğlu | Bülent Sezen

Publisher: İdeal Kent Yayınları

Main Topics: Yazılım Mühendisliği (Diğer)

Journal: OPUS Toplum Araştırmaları Dergisi Vol.22 Issue 3, pp.532-549

Document Type: Article

Publication Date: 5/2025

Document Language: English

Document ID (Org No.): 2001348662

pISSN: 2791-9862

Issues

Authors:

Büşra Aydın | Yavuz Selim Balcıoğlu | Bülent Sezen

Publisher:

İdeal Kent Yayınları

Main Topics:

Yazılım Mühendisliği (Diğer)

Journal:

OPUS Toplum Araştırmaları Dergisi Vol.22 Issue 3, pp.532-549

Document Type:

Article

Publication Date:

5/2025

Document Language:

English

Document ID (Org No.):

2001348662

pISSN:

2791-9862