<![CDATA[Air humidity is a parameter that influences the environment and human activities. Accurate air humidity prediction can be helpful for various purposes, including weather-based decision-making. However, a single model has limitations in capturing non-linear patterns and long-term dependencies in time-series data, making it difficult to predict data well, especially complex and time-series data such as weather. Therefore, a model with a hybrid approach is needed. Hybrid modeling is a combination of two or more learning methods. This study proposes a hybrid approach by combining Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) layers to predict air humidity more accurately and effectively than a single model. CNN is used to extract temporal representations from historical weather data in the form of time sequences, while LSTM is for long-term memory. Prediction is carried out by collecting weather data, data preprocessing, feature transformation (including cyclic feature transformation), designing the CNN-LSTM architecture, model training, and evaluation using evaluation metrics such as mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and coefficient of determination (R²). This study uses weather data from Bogor Regency for the period January 1, 2020 to October 31, 2025 obtained from the Citeko Meteorological Station at coordinates Latitude -6.70000, Longitude 106.85000, and an altitude of 920 meters. The results obtained are that the CNN-LSTM model has an average MAE value of 4.3596, MSE 29.9126, RMSE 5.4689, and R² 0.0756 show that the CNN-LSTM hybrid model is able to improve the accuracy of air humidity prediction compared to a single model.]]>
