A bilevel production planning using machine learning-based customer modeling

Abstract

Mass customization is an important strategy to improve production systems to satisfy customers’ preferences while maintaining production efficiency for mass production. Module production is one of the ways to achieve mass customization, and products are produced by combining modules. In the module production, it becomes much more important for manufacturing companies to reflect customers’ preferences for selling products. The manufacturer can increase its total profit by providing customized products that satisfy customers’ preferences by increasing customers’ satisfaction. In conventional production planning, there are some cases where module production is conducted by the demands from customers’ preferences. However, the customer decision-making model has not been employed in the production planning model. In this paper, a production planning model incorporating customers’ preferences is developed. The customers’ purchasing behavior is generated by using a machine learning model. Customer segmentation is conducted by clustering data that uses the purchase data of multiple customers. The resulting production planning model is a bilevel production planning problem consisting of a single company and multiple customers. Each company can sell products that combine modules that customers require in each segment. We show that the proposed model can obtain higher customers’ satisfaction with greater profits than the model that does not employ the customers’ purchasing model.

Keywords

• supply chain management
• mass customization
• production planning
• customer’s modeling
• machine learning