Improved Data Mining Scheme For Weather Forecasting

Abstract

In big data analytics, weather forecasting is an essential one to identify a possible instance newlinein the future and plan for better water management. Weather forecasting is used in diverse newlineapplications such as climate monitoring, pollution diffusion, drought discovery, weather newlineprediction, agriculture, communication, and so on. During the weather prediction, accurate newlineforecasting is a major concern. Conventional methods were developed to perform weather newlineforecasting with a lower number of features. However, the accuracy and error rate remained newlineunsolved. For that reason, the research work is implemented with three efficient proposed newlinetechniques in order to enhance the performance of future weather prediction with higher newlineaccuracy and lower time and error rate. In this proposed research, the time and error newlinerate issues are solved by means of developing the feature selection process. Besides, the newlineweather condition of a particular location is effectively identified with the implementation of newlineclustering and ensemble classification. newlinePrincipal Component Regression based Iterative Gradient Ascent Expected Maximization newlineClustering (PCR-IGAEM) Model is designed to obtain higher prediction accuracy and newlineminimal time consumption while predicting the weather. Through the Principal Component newlineRegression Analysis (PCRA), the performance of feature selection is improved by means newlineof choosing more essential features for weather prediction. This helps to reduce the newlinecomplexity involved in weather prediction. In addition, the Iterative Gradient Ascent newlineExpected Maximization Clustering (IGAEM) is applied in the proposed PCR-IGAEM to newlineincrease the accuracy of weather prediction. The expected log-likelihood between data and newlinethe cluster center is measured in the clustering process. In IGAEM, gradient ascent is newlineemployed to maximize the likelihood function between the cluster center and weather data newlinefor the grouping process. This, in turn, the more similar data are accurately grouped into a newlineparticular cluster with higher accuracy and less time utilizati