Signi cance of Features via SiZer

  title={Signi cance of Features via SiZer},
  author={J. S. Marron and Parag Chaudhuri},
SiZer is an exploratory data analysis tool, that works in conjuction with smoothing methods. It addresses the issue that is often central to the use of smoothing in data analysis: which observed features are \really there", and which cannot be distinguished from the background sampling variability? The SiZer approach makes heavy use of scale space ideas.