Elastic net feature selected multivariate discriminant mapreduce classification

Abstract

Analyzing the <span>big stream data and other valuable information is a significant task. Several conventional methods are designed to analyze the big stream data. But the scheduling accuracy and time complexity is a significant issue. To resolve, an elastic-net kernelized multivariate discriminant map reduce classification (EKMDMC) is introduced with the novelty of elastic-net regularization-based feature selection and kernelized multivariate fisher Discriminant MapReduce classifier. Initially, the EKMDMC technique executes the feature selection to improve the prediction accuracy using the Elastic-Net regularization method. Elastic-Net regularization method selects relevant features such as central processing unit (CPU) time, memory and bandwidth, energy based on regression function. After selecting relevant features, kernelized multivariate fisher discriminant mapr classifier is used to schedule the tasks to optimize the processing unit. Kernel function is used to find higher similarity of stream data tasks and mean of available classes. Experimental evaluation of proposed EKMDMC technique provides better performance in terms of resource aware predictive scheduling efficiency, false positive rate, scheduling time and memory consumption.</span>

Big data classification based on improved parallel k-nearest neighbor

“…MapReduce is a programming methodology created by Google to handle large-scale data analysis. It is based on the Hadoop framework [11], [58], [78]- [81], [64]- [67], [69]- [71], [74]. It is employed in a wide variety of applications.…”