Machine Learning Based Prediction of Customs Clearance Times in Sri Lanka-An Ensemble Learning Approach

In the global logistics and supply chain industry, delays in customs documentation clearance create significant inefficiencies, reducing productivity. In Sri Lanka, manual processing of customs documentation often leads to prolonged waiting times, negatively impacting stakeholders in the shipment lifecycle. This research develops a machine learning model to predict customs documentation clearance times, aiming to improve transparency and operational efficiency. Using a dataset obtained from relevant authorities, this study incorporates key factors such as origin country, shipment type, and weight for predicting the customs clearance time. Regression algorithms, including Decision Tree and Gradient Boosting, were employed and optimized through a Voting Regressor ensemble approach. The final model achieved high accuracy, with a Mean Absolute Error (MAE) of 0.316, Mean Squared Error (MSE) of 0.16, and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score of 0.98. These results demonstrate the model's potential to provide accurate clearance time predictions, reducing delays and enhancing system productivity.