Taking a look to see how well existing MAF metrics quantify problems with the window function, it appears that the periodigram computed in PeriodicMetric is very undersampled (assuming the 10 year baseline is evaluated). As a result, it will miss most peaks entirely and poorly characterize those that it does observe. I recommend increasing the frequency sampling to oversample 1/baseline by a factor of 5-10. @lundmb
Taking a look to see how well existing MAF metrics quantify problems with the window function, it appears that the periodigram computed in PeriodicMetric is very undersampled (assuming the 10 year baseline is evaluated). As a result, it will miss most peaks entirely and poorly characterize those that it does observe. I recommend increasing the frequency sampling to oversample 1/baseline by a factor of 5-10. @lundmb