12-12-2012, 11:44 AM
Cluster and Calendar based Visualization of Time Series Data
Cluster and Calendar.pdf (Size: 13.8 KB / Downloads: 21)
Abstract
A new method is presented to get insight into univariate time series data. The problem addressed here is how to identify patterns and trends on multiple time scales (days, weeks, seasons) simultaneously. The solution presented is to cluster similar daily data patterns, and to visualize the average patterns as graphs and the corresponding days on a calendar. This presentation provides a quick insight into both standard and exceptional patterns. Furthermore, it is well suited to interactive exploration. Two applications, numbers of employees present and energy consumption, are presented. Time series data are ubiquitous. The aim of time series analysis is to obtain insight into phenomena, to discover repetitive patterns and trends, and to predict the future. We focus here on the analysis of univariate time series data. Suppose, we have collected energy consumption or air pollution data at short time intervals during one year, then how can we extract information from these data? In the next section we discuss the problem and consider various solutions. Current methods fall short in the analysis of time series data at the various time scales, such as years, weeks, and days. Our new approach is based on a combination of two methods: The use of cluster analysis and the visualization of the result on a calendar.