Trend tests in time series with missing values: a case study with imputation

Ramos, Maria do Rosário; Cordeiro, Clara

http://hdl.handle.net/10400.2/14718

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
1.4913055 ICNAAM 2013 Trend Tests in TS with missing Ramos Cordeiro.pdf		513.18 KB	Adobe PDF	Download

Send Feedback

Authors

Ramos, Maria do Rosário

Cordeiro, Clara

Abstract(s)

Testing for trend is an important problem, especially when one is dealing with environmental time series. The tests considered here are the usual t-test and the Mann-Kendall test, a nonparametric version widely used because it requires fewer assumptions. The aim is to assess the performance of two trend tests in time series with autocorrelation after an imputation method is applied to estimate the missing observations. The performance of the trend tests will be illustrated for some well-known data sets existing in R software.

Keywords

Trend tests Mann-Kendall Missing values Sieve bootstrap Time series T-test

URI

http://hdl.handle.net/10400.2/14718

Citation

Ramos, M. R., & Cordeiro, C. (2013). Trend tests in time series with missing values: A case study with imputation. AIP Conference Proceedings 1558, 1909 (2013) https://doi.org/10.1063/1.4825905