Data Mining based Technique for Natural Event Prediction and Disaster Management

Abstract

Due to advancement of technology the computational algorithms and computer based data analysis are used for making large and effective decision. Therefore a significant role on the human life is observed the main aim of developing such kind of technology is to provide ease in various domains and making future planning for managing the risk. Among various issues the disaster is also a critical risk in today’s scenario of India. Thus the risk management and disaster management techniques are developed for keep in track the losses in controlled manner. In this presented work a new model using the data mining techniques for predicting the disaster and their place is proposed for development. Therefore various different data mining techniques and methods are included for developing the accurate and effective data model. The proposed work includes the three main contributions for prediction based technique development. First the preprocessing technique development by which the unstructured data is processed and filtered for transform the information into the structured data format. Therefore in this phase the Bay’s classifier is used, Secondly development of learning technique for accurate pattern learning of the disasters and their places. Therefore in this phase the k-means clustering and hidden Markov model is employed for performing the training. Finally the prediction and their performance evaluation, in this phase the trained model is used to accept the current scenarios and predict the next event.The implementation of the proposed technique is performed using the JAVA technology and for dataset generation the Google search API is used. After the implementation of the proposed system the performance of the system in terms of accuracy, error rate, time complexity and space complexity is evaluated. The experimental results demonstrate the effective and accurate learning of the system. Thus the proposed data model is adoptive and acceptable for the various real world data analysis and decision making task.