Skip to content

grimmlab/Forecasting-Perishables-with-XAI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Forecasting sales and identifying purchase-driving attributes of perishable products with explainable artificial intelligence based on the example of yogurt

This is the official code to our paper: "Forecasting sales and identifying purchase-driving attributes of perishable products with explainable artificial intelligence based on the example of yogurt".

Purpose: Understanding the product attributes that drive consumer purchases is crucial for retailers and marketers to remain competitive. However, extrinsic and intrinsic attributes are often analysed in isolation, overlooking their complex interplay in real-life sales scenarios.

Design/methodology/approach: This study addresses this limitation by evaluating the positive or negative impact of yogurt attributes on sales forecasts, representing properties that support high sales figures of these attributes. Forecasts were generated using the machine learning method eXtreme Gradient Boosting, a decision tree-based ensemble learner. This model was trained on four years of data that included diverse yogurt attributes available to consumers at the point of sale, such as price and nutrient facts. A subsequent Shapley Additive Explanations analysis provided a detailed overview of extrinsic and intrinsic attributes supporting high sales figures.

Findings: EXtreme Gradient Boosting achieved strong forecasting performance with an R^2 score of 0.89. Calorie content emerged as the most important intrinsic attribute, while price was identified as the most influential extrinsic feature.

Practical implications: These findings offer valuable insights for marketers, retailers, and product developers, highlighting the complex interplay of product attributes in shaping yogurt purchase behaviour and showcasing the practical application of machine learning and explainable artificial intelligence in consumer analytics.

Originality: This study offers practical insights into the interpretation of machine learning forecasting models in retail environments. Furthermore, it offers a holistic approach to extrinsic and intrinsic attributes affecting perishable food product purchases.

Contributors

This code is developed and maintained by Josef Eiglsperger, M.Sc. of the Bioinformatics lab lead by Prof. Dr. Dominik Grimm.

Citation

Forecasting sales and identifying purchase-driving attributes of perishable products with explainable artificial intelligence based on the example of yogurt. K Brückner, J Eiglsperger, DG Grimm, K Menrad. British Food Journal, 2025 (under review)

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages