This paper proposes a new hybrid framework called the Time-series Decomposition and Two-stage Attention (TDTA) for the long-term forecasting of container volume. Forecasting the container volume is essential for supporting portfolio decisions about port facility investment plans and port operation plans. This framework includes time-series decomposition to deconstruct time-series into several components (trend, seasonality, and residual), a two-stage attention mechanism that assigns priority to important variables to increase long-term prediction accuracy, and a long short-term memory network that predicts each component and then aggregates all components to derive the final output. In an experiment, the container volume was predicted after six months using the proposed method and compared to the method used in previous studies. The TDTA achieved a better predictive performance than the existing time-series models used in previous studies. Hence, our proposed method can help in decision-making through accurate long-term predictions of container volume, and can also help with the long-term prediction of other time-series data.