Calculation of mean

Why do you have to use @Ceil? @Round seems to be a more logical choice, and it gives 0,43 as a result.

Because of an occasional outlier I can't rely on normal @Round function. The client has a say in this as well, guess they wanna have all rounding to their advantage. And it shouldn't cause any problems if mean is calculated correctly.

But the mean is calculated correctly. 

The difference you get is a result of the inaccuracy of calculating with decimal numbers. So you should @round to a lower number of decimals decimals before applying @ceil, something like:

@ceil(@round(@Value(TheMeanValue),6),2)

@ceil(@round(@Value(TheMeanValue),6),2)

That's what I did as a workaround - but it feels pretty unsatisfying.

And mathematically it is wrong. That can't be discussed. It is calculated correctly only within the bounds of floating point approximations, but that doesn't make it right, it just makes it the best representation. So in this case FME shouldn't perform a calculation at all, it should just transfer the value.

@ceil(@round(@Value(TheMeanValue),6),2)

I don't think FME 'knows' all values are identical. FME simply calculates the mean value from the sum of all values, and the number of values.

But I agree that, when the minumum value equals the maximum value, FME could be instructed to just return that value. 

Maybe you can submit an idea for this. It would help other users who encounter this behaviour.