[liangqin12354]

Dear PHM Challenge committee,

I have a question regarding the PDF. As mentioned in the introduction: “All distributions will be set to a scale = 1 to ensure that the area under the curve is 1 and that the probability of any given value is less than 1. Other distributions found in Scipy can be used upon request.”.

But for PDF like lognorm or beta, even though scale = 1, it is still possible to get score greater than 1.

How should I handle this situation?

Best regards,
Qin

[thejbills]

Our scoring function actually uses the parameters you supply including the scale (I’ll update the description language). But the scoring functions corrects area and peak probability.

If the area is not 1 it’s normalized to 1.
Then
If the max probability is greater than 1, it’s normalized to one.

[Author]

Posts