A machine-learning aid model for estimating lean manufacturing implementation in the 1 3 Latin-American industry

Abstract

Lean Manufacturing is widely recognized as a strategy to improve efficiency, reduce waste, and foster continuous improvement. In Latin America, however, its adoption still varies significantly, and traditional maturity assessments often rely on qualitative surveys that are resource-intensive and difficult to replicate. This article proposes a predictive framework based on supervised machine learning to classify firms according to their level of Lean Manufacturing implementation. The dataset was constructed from closed-ended questions derived from the World Management Survey and refined through expert validation to ensure meaningful labeling of Lean adoption levels. Four ensemble-based classifiers—Random Forest, Gradient Boosting, XGBoost, and LightGBM—were trained and evaluated using accuracy, precision, recall, and F1-score. The results showed that overall accuracy ranged from 0.878 (LightGBM) to 0.891 (Random Forest). Class-level F1-scores were consistently high, reaching 0.96 for the “Low” category and ranging from 0.82 to 0.91 for the intermediate groups. These findings indicate that ensemble methods, particularly Random Forest and XGBoost, provide robust predictive performance while also offering interpretability through feature analysis. The proposed framework provides a scalable and data-driven approach to assessing Lean maturity, supporting decision-making for managers and policymakers seeking to strengthen industrial competitiveness in Latin America.