Fluctuating volumes of Less Than Container Load (LCL) shipments in pharmaceutical packaging create risks of underutilization and overbooking that complicate maritime logistics planning and operational efficiency. This study aims to analyze LCL shipment patterns at PT ABC and to compare the forecasting performance of the Holt–Winters Additive method and a Long Short-Term Memory (LSTM) model in supporting adaptive maritime logistics planning. A mixed-methods approach was applied by combining quantitative time series analysis of monthly operational data from August 2023 to July 2025 with semi-structured interviews with logistics personnel to contextualize the quantitative findings. Forecast accuracy was evaluated using Mean Absolute Percentage Error (MAPE), Mean Absolute Deviation (MAD), and Root Mean Square Error (RMSE). The results indicate that the Holt–Winters model has limitations in handling irregular and highly volatile shipment patterns, producing operationally unrealistic zero-volume forecasts during certain periods. In contrast, the LSTM model generated more stable forecasts, achieving a MAPE of 14.56% and a MAD of 10.38 m³. These findings suggest that LSTM provides greater operational reliability for forecasting LCL shipments under fluctuating conditions. From a practical perspective, more stable forecasts can support a transition from spot-based shipments to contract-based space booking strategies and facilitate consolidation postponement during low-utilization periods to improve freight cost efficiency. Although regulatory compliance is not directly measured, forecasting-informed planning can support logistics practices aligned with Good Distribution Practice (GDP) principles by improving shipment coordination and reducing operational uncertainty.
Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.