Time-varying Data Visual Analysis method based on Parallel Coordinate System

Abstract

With the advent of the era of big data, there are more and more ways to obtain data. The collected data generally is large-scale, multi-dimensional and time-varying. To explore and analyse time-varying data comprehensively, this paper proposes a visual method for processing multidimensional time-varying data based on parallel coordinate system. Firstly, we use MDS algorithm to turn the original multidimensional time data to distribute in one-dimension space in chronological order, constructing the time-varying parallel coordinate system. Then, we cluster the projection coordinate points of each time axis, and each type of data is drawn in different colours to differentiate. At last, edge bundling method is used to cluster data polylines in each time segment, helping to reduce visual confusion and improve information expression efficiency. The experimental analysis of the air quality dataset from Krakow Poland shows that the method can comprehensively analyse the multi-dimensional time-varying data, helping users to explore the implied regularity hidden in the dataset effectively.