Optimized Deep Learning Predictive Model for Food Sales and Demand Forecasting
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
In modern complex business milieu, managing various aspects of the supply chain has become increasingly challenging.It is vital to improve viability, sales, and customer satisfaction by predicting key relational factors.However, traditional forecasting methods often yield inaccurate results and are time-consuming.To solve these issues, CatBoost, a Machine Learning (ML) algorithm, offers advanced features for demand forecasting.Yet, it was computationally intensive when utilizing huge or high-dimensional data with numerous unique categorical values.Also, parameter tuning can be complex.Thus, this paper proposes an optimized Deep Learning (DL) model to predict food sales, profit, and delivery times using ecommerce dataset.First, the raw data from food supply chain-related business logs is collected and pre-processed to handle missing values and outliers.Then, a Bidirectional Long Short-Term Memory (BiLSTM) network is employed for demand prediction.Besides, the Binary Coati Optimization Algorithm (BCOA) is applied to optimize the BiLSTM hyperparameters.Finally, experimental results show that BCOA-BiLSTM outperforms traditional ML algorithms in Supply Chain Management (SCM) with a minimum forecasting error.
Fields of Study & Subjects
Forecasting Techniques and Applications
Stock Market Forecasting Methods
Supply Chain and Inventory Management
Publication details
- Journal / source
- Advances in intelligent systems research/Advances in Intelligent Systems Research
- Pages
- 1177–1188
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Book chapter