Uso da análise de dados e Machine Learning aplicado ao processo de separação e expedição em home centers: estudo de caso

Abstract

The order picking and customer service process in the expedition area of home centers presents high operational variability, resulting in delays in meeting internal service targets and directly affecting customer experience. Factors such as total load weight, number of items, delivery type, and operator productivity contribute to fluctuations in separation and service times, making it difficult to efficiently control operational flow. In this context, it becomes necessary to quantitatively understand the factors that influence delays and propose data-driven improvements. This Final Term Paper aims to analyze the separation and service process in the expedition area, identifying operational bottlenecks and proposing solutions through statistical analysis and Artificial Intelligence techniques. A realistically structured fictitious database containing 3,000 simulated records was developed according to the company’s operational rules. Descriptive statistical analyses were conducted to identify patterns, variability, and compliance with internal service targets. Additionally, supervised classification models, such as Decision Tree and Random Forest, were applied to predict service delays. The results showed that approximately 40% of orders exceeded the established service-time targets. Feature importance analysis indicated that total weight, operator productivity, and number of items are the main factors associated with delays. The study concludes that integrating process engineering, statistics, and Artificial Intelligence is an effective approach to support logistics optimization and improve operational decision-making.