Multi-Granular Trend Detection for Time-Series Analysis

Abstract

This is the accepted version of the paper.This version of the publication may differ from the final published version. Abstract-Time series (such as stock prices) and ensembles (such as model runs for weather forecasts) are two important types of one-dimensional time-varying data. Such data is readily available in large quantities but visual analysis of the raw data quickly becomes infeasible, even for moderately sized data sets. Trend detection is an effective way to simplify time-varying data and to summarize salient information for visual display and interactive analysis. We propose a geometric model for trend-detection in onedimensional time-varying data, inspired by topological grouping structures for moving objects in two-or higher-dimensional space. Our model gives provable guarantees on the trends detected and uses three natural parameters: granularity, support-size, and duration. These parameters can be changed on-demand. Our system also supports a variety of selection brushes and a time-sweep to facilitate refined searches and interactive visualization of (sub-)trends. We explore different visual styles and interactions through which trends, their persistence, and evolution can be explored. Permanent repository link

Uncertainty-Aware Seasonal-Trend Decomposition Based on Loess

“…Visual analytics for time-oriented data has employed seasonal-trend decomposition [21] and also uncertainty quantification and visualization [22]. Visualization methods that combine aspects of time-varying data and uncertainty commonly work with time series ensembles, such as the works by Ferstl et al [23] and Van Goethem et al [24]. We also consider such ensembles as a source of uncertainty.…”