Comprehensive Survey of Classification, Streaming Techniques in Big Data Analytics

Abstract

Now a days, managing big data are a practical issue because of large number of data to be increased data by day from some real time applications like web clicks log data, social media data, mobile communications data, customer service center data, e-commerce related web data and health care sensor data. Basic nature of big data is structural but some of generated data may have some inconsistent, inaccessible and unstructured. Big data streaming is a process in which big data is quickly processed in order to extract real-time insights from it. So that traditionally some of the streaming techniques are introduced to process inaccessible data and visualize that data and represent in various patterns. In this paper, we present the comprehensive survey/study of data streaming techniques with respect to visualize and process continuously generated data from embedded sensors, some IoT devices, real time transactions and mobile applications. We also describe the brief discussion about different prediction related classification and clustering approaches used for data processing with similar or dissimilar patterns. We give brief comparison of different approaches used in visualization/processing of big data with different feature representations. Also presents state-of-the-art methodologies relates to anomaly detection and also describe vital characteristics associated some machine learning related algorithms.