March 17, 2022
An R Shiny app for trend testing using the Mann-Kendall test and estimation of trend magnitude using the Theil-Sen slope estimator. These methods are suited for univariate time-series with monotonous trends and no seasonal or other cycles in the data (no autocorrelation in the time-series). Nonparametric methods are preferred to parametric methods (i.e,. ordinary linear regression analysis) because they make no assumption about the probability distribution of the data. Additionally, nonparametric methods and, in particular methods for the estimation of trend magnitude, are robust to the presence of outliers or to abrupt breaks due to inhomogeneous time-series.
This app was created by Charles Holbert. Check out the app here. The User Guide section of the app contains more information.
- Posted on:
- March 17, 2022
- 1 minute read, 119 words