作者: Weiyi Xie , Sebastian Kurtek , Karthik Bharath , Ying Sun
DOI: 10.1080/01621459.2016.1256813
关键词:
摘要: ABSTRACTWe propose a new method for the construction and visualization of boxplot-type displays functional data. We use recent data analysis framework, based on representation functions called square-root slope functions, to decompose observed variation in into three main components: amplitude, phase, vertical translation. then construct separate each component, using geometry metric space, novel definition median, two quartiles, extreme observations. The outlyingness is very complex concept. Thus, we identify outliers any components after decomposition. provide variety tools proposed including surface plots. evaluate extensive simulations focus our attention real applications exploratory sea tempe...