Rainfall prediction is crucial for agriculture, water resource management, and disaster preparedness. This study introduces a rainfall prediction approach by hybridizing a Convolutional Neural Network (CNN) front-end with a Long Short-Term Memory (LSTM) back-end. The proposed model captures the short-term patterns and long-term dependencies in rainfall data, addressing rainfall’s complex, non-linear nature. The study utilizes historical rainfall data from the Tiruvangadi region of Kerala, India, covering 98 years (1918–2016) and collected from 116 weather stations. The data was pre-processed and transformed into a binary classification task, predicting the presence or absence of rainfall on a given day. The CNN-LSTM model employs a 1D convolutional layer followed by LSTM layers, incorporating techniques like L2 regularization and dropout to prevent overfitting issues. The proposed hybrid model achieved test accuracy of 82.53%, along with balanced precision and recall scores of 0.8253 and 0.8253, respectively, depicting reliability in forecasting rainy and non-rainy days. This study underscores the potential of deep learning (DL) models to improve rainfall prediction accuracy.