Data visualization for functional time series

11 December 2018

Presentation to MATRIX Workshop on Functional Data Analysis and Beyond

Any good data analysis begins with a careful graphical exploration of the observed data. For functional time series data, this area of statistical analysis has been largely neglected. I will look at the tools that are available such as rainbow plots and functional box plots, and propose several new tools including functional ACF plots, functional season plots, calendar plots, and embedded pairwise distance plots. These will be illustrated using pedestrian count data in Melbourne, smart meter data from Ireland, and mortality data from France.

  1. Using ggplot2 for functional time series. R code
  2. Time-indexed probability distributions (slides below)
 Download pdf

« Seasonal functional autoregressive models | Feature-based forecasting algorithms for large collections of time series »