What type of study is this?

This is a Quantitative Study study.

September 18, 20250 citationsOpen Access

Flood Prediction with Artificial Intelligence An Exploratory Data Analysis Approach

AMAnna Mané RHRashmi Ravindra Halkarni PBP. N. Bhat

Key Points

The LSTM network achieved an accuracy of 98.72% in predicting flood occurrences, indicating its effectiveness compared to other models.
Environmental and hydrological data were analyzed to predict floods on both national and regional levels, emphasizing the study's broad applicability.
Machine learning and deep learning algorithms were employed, enhancing the predictive capabilities for flood forecasting.
Bootstrap Forest was rejected for its overfitting issues, while the LSTM model emerged as the superior choice for flood prediction.

Abstract

Abstract This paper presents the application of various ML and DL algorithms to de- termine occurrence of floods in India. In this study data driven methods are used which could help in predicting floods on a national and regional level. Using a real- world dataset containing environmental parameters (Minimum Temperature, Maximum Temperature, Minimum pH, Maximum pH, Rainfall level) and hydro- logical data (Reservoir level, Live capacity, Storage capacity, Current water level), the trends are analyzed, detect anomalies, and identify attributes influencing flood occurrence. A wide range of predictive models including ML techniques like Boot- strap Forest,, Support Vector Machine, Naive Bayes, KNN, ANFIS, and advanced DL models like ANN and LSTM networks. Boostrap Forest delivered an accuracy of 99.43was rejected due to its overfitting nature ,Hence LSTM Network was con- sidered the best and suitable model with an accuracy of 98.72%. Keywords: AI, RNN, ANFIS, GMDH, Flood prediction

Ask AI

Helpful

Bookmark

View Full Paper