Integrating time-series visualizations within parallel coordinates for exploratory analysis of incident databases

Aviation disaster prevention has always been of interest to homeland security, especially after the recent use of aircrafts as weapons by terrorists. With better understanding of the deficiency of different types of aircraft and their corresponding effects on the craft's safety, better maintenance and response plans can be devised to prevent disasters from occurring. In this paper, we present a visual analytical technique to examine the Federal Aviation Agency's Accident/Incident Database, which contains more than 90,000 incidents across 53 dimensions over the last 30 years, for identifying trends of relationships between dimensions over time. Our technique is based on the integration of the ThemeRiver technique directly within a parallel coordinates framework, and simultaneously presents both a "forward flow view" and a "backward flow view" between each dimension. The forward flow view shows the trends over time of each of the elements-of-interest in the first dimension, while the backward flow view illustrates how the elements in the second dimension contribute to the overall trends seen in the first dimension. Through the use of our technique, we were able to identify characteristics of aircrafts and suggest plausible explanations to their common failures.