Kendall's Tau as the test for trend in time series data
2012
Cabilio, Paul Zhang, Ying
Acadia University
Master of Science
Masters
Mathematics and Statistics
Mathematics & Statistics
When using nonparametric tests test time series data for trends, most require that the data the model independent and identically distributed. course, real-world data not always meet this requirement. For instance, the case auto regressive process, the successive observations the time series are not stochastically independent. result the usual Kendall's tau test not valid. This study will discuss the ect correlated error signicance levels and power Kendall's tau, the asymptotic distributions U-statistics for stationary ARMA process and the circular Bootstrap method. Modi cations these tests will developed test for trends non-independent stationary models. The signicance level and power for the Bootstrapping and Monte Carlo methods will simulated using the statistical software package version 2.9.2. Finally, these methods will applied example.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.
https://scholar.acadiau.ca/islandora/object/theses:222