Simulation-based modules for teaching data-driven decisions in industrial engineering

Martínez J.L. · IISE Annual Conference and Expo 2026 (2026) · DOI: 10.21872/2026iise_17642

Resumen

Industrial engineering students often learn analytics algorithms in isolation from the managerial decisions they are meant to inform. This paper describes the design and development of interactive, simulation-based modules that connect core data-mining techniques to end-to-end decision making in a retail-banking context. The flow comprises customer profiling (PCA and K-means), selection/scoring (logistic regression with threshold control), pricing analysis (multiple linear regression), and early-risk monitoring (multiclass decision trees). A reference implementation in R/Shiny was produced using a generative-AI-assisted development workflow for both frontend and backend, accelerating prototyping, documentation, and code refactoring. Each learner interacts with a unique, synthetic dataset generated from fitted distributions to obtain unique yet reproducible scenarios. The interface emphasizes decision parameters and provides immediate visual feedback to foster interpretation, hypothesis-driven exploration, and an understanding of realistic business trade-offs. By foregrounding the connection between model tuning and managerial action, this work shifts IE education from method-centric instruction to decision-centered learning. The accompanying toolkit-comprising the application, instructor guides, and student guides-lowers adoption barriers and establishes a shared vocabulary for key concepts like risk thresholds and pricing trade-offs, making it suitable for analytics, operations, and engineering-management courses. © (2026) by Institute of Industrial & Systems Engineers (IISE) All rights reserved.

Instituciones: Pontificia Universidad Católica del Perú