227 – Clustered Functional Data Analysis
Graphical Representation of Clustered Functional Actigraphy Data
Abbass Sharif
Utah State University
Juergen Symanzik
Utah State University
Human activity and rest patterns could be measured through a device called an actigraph, which collects measurements almost continuously over time. These actigraphy measurements are often treated as functional data. In this article, we first introduce four techniques to visually detect clusters in raw functional actigraphy data for patients with different depression levels. These techniques are: (i) density-based plots, (ii) data enveloping methods, (iii) data summing over time, and (iv) multivariate time series plots. No filtering, smoothing, or any other statistical techniques are required to implement these visualizations. Then, we provide a brief glance at the underlying implementation of these techniques in the R package ActiVis. We finish with a brief description of the graphical user interface for this R package.
