Author(s)

INTAEK JUNG, KYUSOO CHONG

Abstract

The Korea Institute of Construction Technology (KICT) is currently developing a platform system for the analysis of driving environment on the road using vehicle sensing data and various public data. The public data in the KICT platform have different collection cycles and space units to provide real-time road weather and traffic information to users. So, it is difficult to build a database in the form of a relational database to use as explanatory variables for driving environment prediction. This study aims to suggest effective time-series data interpolation and spatial data matching to build an integrated database using different types of public data. We suggested three interpolation methods (piecewise constant interpolation, linear interpolation, nonlinear interpolation) using time-series data and the spatial data matching using weighted average based on each administrative boundary. The quadratic and cubic spline interpolation was applied as the nonlinear interpolation method in this study. As a result of the case study, the linear spline interpolation was selected as the best method and the spatial matching method also showed good results. Although the linear spline interpolation is not so different from other interpolation methods in terms of estimation error, the linear spline interpolation is more effective than other interpolation methods in terms of algorithm implementation and computation speed to be applied to the system. We expect that the built database will be widely used as an explanatory variable for developing various driving environment prediction models.

Keywords

interpolation, spatial matching, database, public data, vehicle sensing data