Reproduce “Forecasting sales with Bayesian networks: a case study of a supermarket product in the presence of promotions” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Muhammad Hamza, Mahdi Abolghasemi, Abraham Oshni Alvandi
Researchers from Monash University developed a Bayesian Network model to forecast supermarket product sales, specifically addressing the complexities introduced by promotions. The model achieved a 1% Mean Absolute Percentage Error (MAPE) for overall weekly sales, outperforming the retailer's existing method by 2 percentage points, and demonstrated capabilities for diagnostic inference.
Key insights
- Bayesian Networks are uniquely suited for promotional sales forecasting as they can seamlessly integrate diverse quantitative and qualitative factors within a coherent causal framework.
- The graphical representation of BNs provides interpretability, allowing managers to visualize and understand the causal relationships between factors and their impact on sales.
- BNs offer a method for robust forecasting even in data-scarce scenarios, as expert knowledge can be leveraged to elicit prior probabilities, a common hurdle for purely data-driven models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.