MWMK Test - rileywheadon/ffa-framework GitHub Wiki
The MW-MK Test is used to identify a statistically significant monotonic trend in the variances of an AMS time series.
- Null hypothesis: There is no significant trend in the variance of the AMS.
- Alternative hypothesis: There is a significant trend in the variance of the AMS.
To compute the AMS variances we use a moving window:
- Set the length of the moving window
$w$ and the step size$s$ . - Compute the standard deviation over indices
$[1, w]$ . - Move the window forward by
$s$ . - Check if the right boundary is beyond the end of the data. If not, go to (5).
- Compute the standard deviation again. Return to (3).
Then, we perform the Mann-Kendall Test on the time series of variances.
For more information about the Mann-Kendall test, read the documentation or run the command Rscript stats.R -n mk
with include_details = TRUE
in config.yml
.