Comparative Analysis of Deep Learning Algorithm for Flood Prediction Probability With and Without Feature Selection

This study investigates the application of deep learning algorithms in predicting flood probability, with a focus on comparing the performance of Long Short-Term Memory (LSTM), Convolutional Neural Networks (CNN), and Backpropagation Neural Networks (BPNN) both with and without feature selection. Utilizing the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology, the research employs a dataset of 21 columns and 50,000 rows sourced from Kaggle. The findings highlight the strengths and limitations of each model, demonstrating that while LSTM provides the highest accuracy, CNN and BPNN offer faster processing times. The comparative analysis underscores the importance of model selection in flood prediction tasks, particularly when balancing the trade-offs between accuracy and processing efficiency. The results suggest that the choice of model should be aligned with the specific needs of the application, whether prioritizing accuracy or processing speed.