This study presents a novel approach combining LSTM and XGBoost algorithms to predict storm characteristics and occurrence in Western France. Storms pose significant risks to human life and the environment, with increasing frequency and severity. The research uses data from buoys and a storm database (1996–2020) to train and test models. The LSTM model accurately predicted temperature and pressure but struggled with extreme wave heights and wind speeds. The XGBoost model, however, showed excellent performance in predicting storm occurrence. The study combines both models to forecast storm characteristics and occurrence, using LSTM for regression-based predictions and XGBoost for classification. The models were trained on data from 1996–2015 and tested on 2016–2020. The LSTM model predicted temperature and pressure with high accuracy, while humidity, wind speed, and wave characteristics showed lower accuracy. The XGBoost model achieved perfect recall and specificity, accurately predicting all storms in the test period. The study highlights the effectiveness of LSTM and XGBoost in storm prediction, offering a comprehensive method for forecasting storm characteristics and occurrence. The results demonstrate the potential of these models in improving storm preparedness and risk reduction strategies. The study also emphasizes the importance of using diverse storm variables and addressing data imbalance in storm prediction. The findings contribute to the field of extreme weather prediction by integrating LSTM and XGBoost for improved accuracy and reliability in forecasting storm events.This study presents a novel approach combining LSTM and XGBoost algorithms to predict storm characteristics and occurrence in Western France. Storms pose significant risks to human life and the environment, with increasing frequency and severity. The research uses data from buoys and a storm database (1996–2020) to train and test models. The LSTM model accurately predicted temperature and pressure but struggled with extreme wave heights and wind speeds. The XGBoost model, however, showed excellent performance in predicting storm occurrence. The study combines both models to forecast storm characteristics and occurrence, using LSTM for regression-based predictions and XGBoost for classification. The models were trained on data from 1996–2015 and tested on 2016–2020. The LSTM model predicted temperature and pressure with high accuracy, while humidity, wind speed, and wave characteristics showed lower accuracy. The XGBoost model achieved perfect recall and specificity, accurately predicting all storms in the test period. The study highlights the effectiveness of LSTM and XGBoost in storm prediction, offering a comprehensive method for forecasting storm characteristics and occurrence. The results demonstrate the potential of these models in improving storm preparedness and risk reduction strategies. The study also emphasizes the importance of using diverse storm variables and addressing data imbalance in storm prediction. The findings contribute to the field of extreme weather prediction by integrating LSTM and XGBoost for improved accuracy and reliability in forecasting storm events.