<![CDATA[Dynamic changes in air temperature in the airport environment significantly affect the safety and efficiency of flight operations, particularly in determining aircraft load restrictions and calculating runway length requirements during takeoff. Therefore, precise daily air temperature forecasting at Juanda International Airport is crucial to support flight safety. This study aims to model and forecast daily average air temperature using a Machine Learning approach with the Random Forest Regression algorithm. This algorithm was chosen because of its superiority in handling complex historical time series data through pre-processing and feature engineering stages. Temperature fluctuation patterns are extracted using the sliding window method to produce lag features and rolling mean as predictor variables that allow the model to capture data characteristics non-linearly. Model performance evaluation is carried out using the Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE) metrics. The test results show that the Random Forest Regression algorithm has excellent performance with an initial prediction error rate (MAPE) of 2.27%. Through the implementation of hyperparameter optimization, model accuracy was successfully improved, reducing the MAPE value to 2.14%. This high level of accuracy demonstrates the model's reliability and consistency in predicting temperature variability at the study site. These forecasting results are expected to be used as a decision-support tool for airport authorities in operational management and mitigating aviation weather-related risks.]]>
