Forecast Combination for Short Container-Throughput Records: Constituent Specification, Gating and Weighting for Port Resource Planning

Monthly container-throughput forecasts inform port resource planning, but short records spanning periods of disruption complicate the choice of forecast. We evaluate the specification, gating and weighting of SARIMA and LSTM forecasts on four monthly series (January 2021 to November 2025) using twelve rolling-origin folds, block-bootstrap intervals, sign tests and scaled errors; with four series, this is a proof-of-concept evaluation, and all comparisons are hypothesis-generating. The mean absolute percentage error of SARIMA moved by 2.77 points across seasonal-differencing specifications, whereas combination rules differed by at most 0.72 points. Under the primary specification, which sets seasonal differencing from training-window seasonal strength, the simple average recorded 8.96%, and no gate or weighting rule was distinguishable from it. An earlier inverse-error weighting advantage did not persist. A five-model pool supported neither removal nor down-weighting of constituents. LSTM variants with deseasonalised input or cross-series training did not improve panel accuracy. Re-forecasting monthly, the average beat seasonal naive (6.86% against 9.62% one month ahead). Forecasts under-planned on most folds; cost-quantile plans, not bias correction, lowered loss at a 5:1 cost ratio. At Terminal A, perfect throughput foresight would still leave a 16% error in crane hours, which limits resource claims. Energy and emissions were not measured.

Authors

Institutions

Publication Details

Journal
Sustainability
Published
2026-11-18
DOI
https://doi.org/10.3390/su182211765
Primary Topic
Forecasting Techniques and Applications
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Forecast Combination for Short Container-Throughput Records: Constituent Specification, Gating and Weighting for Port Resource Planning

Jung-Sik Jeong, Yeonwon Kim, Dong Kwan Kim, Young-Wuk Lee
2 links
Sustainability
Forecasting Techniques and Applications
article

Forecast Combination for Short Container-Throughput Records: Constituent Specification, Gating and Weighting for Port Resource Planning

Jung-Sik Jeong, Yeonwon Kim, Dong Kwan Kim, Young-Wuk Lee
article en

Abstract

Monthly container-throughput forecasts inform port resource planning, but short records spanning periods of disruption complicate the choice of forecast. We evaluate the specification, gating and weighting of SARIMA and LSTM forecasts on four monthly series (January 2021 to November 2025) using twelve rolling-origin folds, block-bootstrap intervals, sign tests and scaled errors; with four series, this is a proof-of-concept evaluation, and all comparisons are hypothesis-generating. The mean absolute percentage error of SARIMA moved by 2.77 points across seasonal-differencing specifications, whereas combination rules differed by at most 0.72 points. Under the primary specification, which sets seasonal differencing from training-window seasonal strength, the simple average recorded 8.96%, and no gate or weighting rule was distinguishable from it. An earlier inverse-error weighting advantage did not persist. A five-model pool supported neither removal nor down-weighting of constituents. LSTM variants with deseasonalised input or cross-series training did not improve panel accuracy. Re-forecasting monthly, the average beat seasonal naive (6.86% against 9.62% one month ahead). Forecasts under-planned on most folds; cost-quantile plans, not bias correction, lowered loss at a 5:1 cost ratio. At Terminal A, perfect throughput foresight would still leave a 16% error in crane hours, which limits resource claims. Energy and emissions were not measured.

SustainabilityVol. 18(22)
Mokpo National Maritime University (KR)
Ministry of Education - Singapore
Industry, innovation and infrastructure, Climate action
Forecasting Techniques and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon