Fast Mode Regression in Big Data Analysis

Abstract

Grasping hidden patterns and unknown correlations of big data quickly and precisely is required in our time. Developing methodology for big data is the focus of big data analysis. Methods do not need to be expensive but meaningful and easy to implement. This paper proposes a mode-based fast pattern grasping method and mode-based fully parametric regression for big data analysis. By taking more than a decade of the 'Health Survey for England' data as an example, we apply the method to uncover the dependency of a response variable on other factors. The dependency relation can be used to check the effect of covariances on the pattern and make pattern or trend prediction easily and quickly.