Negative values for cells that shouldn't be negative
mesotron opened this issue · 2 comments
S1L is a lognormal distribution ranging from .0001 to .1. Therefore it should be impossible for 1 - (1 - S1L)^10 to be negative, but this produces highly negative values. Is this a bug, or am I just missing something basic? See https://www.getguesstimate.com/models/13434 .
When it says "0.0001 to 0.1", that's the 90th percentiles; some numbers are outside that range.
What's happening here is that there are 1-2 outliers or so that are over 1.0 I think, then when they are raised to the ^10 power get pretty huge.
One way around this is with the =filterGreaterThan command. You can do =filterGreaterThan(S1L,1) or similar to remove the few outliers greater than 1.
Looks like that's exactly it. Thanks so much for your help.