Productivity Improvements in Manufacturing Industries Using Machine Learning Algorithm

Abstract

This study combines Lean-Kaizen manufacturing principles with advanced machine learning, specifically using a Random Forest Regressor, to enhance productivity in a manufacturing environment. By applying 5S practices and leveraging data from various operational factors, the integration of machine learning helped optimize key processes such as inventory management, predictive maintenance, and scrap reduction. The Random Forest model identified manual packaging time, number of custom pallets produced, and maintenance downtime as the most significant drivers of productivity, with feature importance scores of 0.22, 0.21, and 0.17, respectively. Machine downtime was reduced by 30%, while maintenance costs decreased by 25%, leading to an overall productivity increase of 15-20%. This quantified approach demonstrates that integrating machine learning into traditional Kaizen methodologies results in sustained improvements in operational efficiency and output.