Hybrid Machine Learning and Lean-S&OP Framework for Logistics Performance Improvement in Peruvian Manufacturing SMEs

Small and medium textile manufacturers in Peru face persistent logistics constraints that reduce competitiveness, including low on-time delivery, poor forecast accuracy, and excessive inventory levels. This study proposes a phased improvement framework integrating machine learning demand forecasting, lean manufacturing practices, and Sales & Operations Planning (S&OP). The framework was developed with five Peruvian textile SMEs and evaluated through discrete-event simulation using calibrated historical data. Results indicate improvements in on-time delivery (from 72% to 97%), inventory turnover (from 4.2 to 6.8 cycles/year), forecast error reduction (MAPE from 18.4% to 8.7%), production lead time reduction (from 12.5 to 8.1 days), and logistics cost reduction (18.3%). Required investment ranges from USD 35,000 to USD 55,000 over 5-6 months. Findings suggest that resource-constrained manufacturers can achieve gradual progress toward Industry 4.0 adoption through staged, affordable, and operationally integrated transformation.