The big data analysis of rail equipment accidents based on the maximal information coefficient

Abstract

With more electrical and electronic equipment applied into the railway system, much more data can be collected and then the big data era of railway is coming. By employing the maximal information coefficient (MIC), the big data analysis of rail equipment accidents is studied to investigate the effect of the updating of rail equipment. The rail equipment accident data set of 25 years (from 1990 to 2014) is separated into three subsets corresponding to the period of the occurrence time of accidents. For every subset, the contributing factors to accident damage, to accident severity, and to accident cause are analyzed, respectively. The results show that the variation trend of the number of rail equipment accidents is more consistent with the variety of railroad service miles rather than carloads. And the factor of highway-rail grade crossings is an important one which accords with the facts. However, a seemingly surprising result is found that there will be more contributing factors to accident severity and to accident causes with more equipment applied into the railway system as time goes on.

Keywords:

• Big data
• rail equipment accident
• the maximal information coefficient
• temporal analysis

Acknowledgments

The first author would like to thank Dr. Laurence R. Rilett, from Nebraska Transportation Center in University of Nebraska-Lincoln. However, we are solely responsible for all views and analysis presented in this paper.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (Grant No. 11671235) and Shandong Provincial Natural Science Foundation, China (Grant No. ZR2018MG003).