Closed rnugent3 closed 4 days ago
And it looks to me like the mean is not the mean, the mean is actually the median? The object created at the below location which appears to be labeled mean in the user interface uses the default probability argument of 0.5
@Brennan1994 would you take a look at this, please?
It could be that if this data is referenced in decision documents, we need to run the full on parametric bootstrap for the uncertainty.
I have pulled this data from the IDE on the input function in the screen shot above Frequency | Flow | Stage |
---|---|---|
0.5 | 202306 | 23.86 |
0.2 | 373981 | 27.38 |
0.1 | 508002 | 29.5 |
0.04 | 696277 | 31.66 |
0.02 | 848335 | 33.04 |
0.01 | 1009190 | 34.58 |
0.005 | 1179084 | 36.11 |
0.002 | 1417793 | 37.58 |
If I increase the number of realizations from 10 to 100,000, the median function (labeled mean in the image) looks like this
Increasing realizations to 100,000 does not slow down the user interface much, but it also does not get us to the exact input function.
500,000 realizations do not really improve the function from 100,000 realizations
The analytical flow-frequency function user interface does not display the input function, only statistics of a very quick (not full compute version) parametric bootstrap about the input function (0.05, Mean, 0.95), see image below.
If our little parametric bootstrap is good enough, then the median "computed" curve should match the input function and we should change mean to median.
If not, then we need to consider redesigning the user interface.