Main Article Content

Sumita Mukherjee, et. al.


The regression analysis, time series, logistic models, neural networks, the Bayesian belief network, and decision trees, are associated with big data, data mining research, algorithm not only to help forecasting but also used for prediction of earthquake. Seismic wave propagation in the form of the earth's crust, is responsible for earthquake occurrences and depends on associated variables and is to be determined from records obtained and received form Nepal Meteorology Department having different factors like depth, magnitude, location, latitude, longitude etc.  by mining methods and results are then evaluated thoroughly. This paper focuses on implication of seismic pattern, trends, association, comparison of earthquake occurrences on statistical data, using time series mining. This paper also studies and correlates various factors with the seismic activity for predicting occurrences of earthquake by developing machine learning models through visualizing time series pattern. It recognizes a strong pattern and orchestrates earthquake prediction. The machine learning method used Python programming to generate accurate graphs neural networks modelling and explains overfitting, stationarity and parsimony features for earthquake prediction.  Data is in a non-linear relationship uses curved fitting regression.  In this study, the comparative results of seismic time series are analyzed at testing point and prediction point. To keep down the destruction by earthquake expert decision systems by Neural network can be developed only using seismic time series analysis having different factors which can be a good study and explanation to develop an algorithm using the following methodology Auto regression (AR), Moving Average (MA), Seasonal Autoregressive Integrated Moving-Average (SARIMA).

Article Details