Performance Analysis on Decision Tree and SVM Classification With Reference to Flood Occurrences in India

Abstract

Floods are pervasive catastrophic disasters which cause financial damages, lost life and environmental scarcity. Flood damage estimation is one of the important factors to the way in the depth of the flood and predicts the future damages. Data mining classification techniques to discover patterns and sequences will use to predict the zones that depiction to flood. A data of flood damages for twelve years collected from various sources. The classification models of Decision Trees and Support Vector Machine with different kernel functions are taken for the prediction models. The Matlab analytics compare the prediction models indicates the performance of the algorithm has much better accuracy. Experiments confirm that the Support Vector Machine with Quadratic kernel function is more accurate in finding the prediction pattern. The accuracy and visualization also suggest that flood prediction which the outcome will goal to better manage floods all the way through preclusion, fortification and catastrophe.

Prediction of Seismic zone in India using Neural Network Algorithms

“…developed a model using support vector machine with different kernel functions to predict the catastrophe flood [5]. Emad Amar et al, (2014) have recommended the neural network system extracts the data through learning process and analyzes nonlinear structures.…”