<![CDATA[Air temperature prediction is important for supporting weather monitoring and operational activities in port areas. Accurate temperature forecasting can support decision-making related to maritime transportation, logistics, and weather-based risk mitigation. This study compares the performance of Long Short-Term Memory (LSTM) and Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) models for daily air temperature prediction in the Tanjung Priok Port area. The dataset consists of daily meteorological observations collected from the Tanjung Priok Maritime Meteorological Station, Indonesia, covering the period from 2000 to 2025. Eight input variables were used, including rainfall, sunshine duration, air pressure, average humidity, average wind speed, and three lagged temperature variables. Data preprocessing included data cleaning, 7-day moving average smoothing, MinMaxScaler normalization, and sequence generation using a sliding-window approach. The dataset was divided into training (70%), validation (15%), and testing (15%) sets. Model performance was evaluated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²). The results show that the LSTM model achieved the best overall performance, with an MAE of 0.1457 °C, RMSE of 0.1811 °C, MAPE of 0.5030%, and R² of 0.9423. In comparison, the CNN–LSTM model obtained an MAE of 0.2566 °C, RMSE of 0.3316 °C, MAPE of 0.8802%, and R² of 0.8065. These results indicate that the standalone LSTM model performed better than the CNN–LSTM model in predicting daily air temperature for the Tanjung Priok Port dataset.]]>
