This paper presents an approach to analyzing two-dimensional temporal datasets focusing on identifying observations that are significant in calculating the outliers of a scatterplot. We also propose a prototype, called Outliagnostics, to guide users when interactively exploring abnormalities in large time series. Instead of focusing on detecting outliers at each time point, we monitor and display the discrepant temporal signatures of each data entry concerning the overall distributions. Our prototype is designed to handle these tasks in parallel to improve performance. To highlight the benefits and performance of our approach, we illustrate and validate the use of Outliagnostics on real-world datasets of various sizes in different parallelism configurations. This work also discusses how to extend these ideas to handle time series with a higher number of dimensions and provides a prototype for this type of datasets.
updated: Wed Oct 30 2019 04:24:38 GMT+0000 (UTC)
published: Wed Oct 30 2019 04:24:38 GMT+0000 (UTC)