A Normalized ANN Model for Earthquake Estimation

Earthquake is one of the most devastating natural catastrophes, mainly because there is rarely any advance notice and hence little opportunity to react. This makes the issue of earthquake prediction highly crucial for human safety. This paper offers a technique for predicting earthquakes by using normalized artificial neural network (ANN). Exploratory Data Analysis (EDA) is applied on the raw dataset to find outliers and the co-relationship between input features. Then, Feature Engineering is performed to normalize the data and remove all unnecessary features. The training data is fed into the neural network model, which generates certain output. Error is calculated based on actual and generated output. Backpropagation algorithm is applied to minimize the error, after which this data is used to train the model. Finally, the Testing data is fed into the model to calculate accuracy and other performance metrics. The outcomes of several experiments are promising. Accuracy of prediction obtained was 94.3%. Also, the training and testing delay recorded were 2.12 seconds and 3.14 seconds respectively.