Data science: Fail of prediction algorithm?


I have a question regarding the following exercise of the Data Science curriculum: Exercise about a prediction algorithm

I tried the values 220 for the flipper and 35 for the bill. It predicted that this penguin would belong to the Gentoo species.

Isn’t that a wrong prediction? My intuition would be that one should compare the mean length and then compare the deviation from the mean. The deviations for the flipper are approximately 0 for Gentoo and 30 for Adelie when one uses my value of 220. The deviation for the bill is approximately 0 for Adelie and 15 for Gentoo when using the value of 35. So far it would mean that Gentoo has a lower deviation from the two values of bill and flipper.

But there is one point that isn’t considered in my view. Flippers are less long than bills. This means that in terms of anatomy a difference of 30 for the flipper is less important than a difference of 15 for the bill. Flippers are usually longer than bills. That means that a small difference in the length of the bill is of greater importance than a big difference in the length of the flipper.

I believe that it would be more appropriate to predict that a penguin with the values 220 for the flipper and 35 for the bill is more likely to be an Adelie penguin instead of a Gentoo. Do you agree?

I looked at this too and had wondered the very same thing. I think the prediction in the example is wrong. It should be Adelie rather than Gentoo penguin. Maybe I’m missing something here(?)
Did you plug in different numbers? Do you agree with the prediction(s) or did you find they were off?

I believe the algorithm isn’t working properly. I made another attempt with the values 200 for flipper and 35 for the bill. It says that it would be a Chinstrap species. This is obviously wrong, because it would fit much better to the Adelie cluster. I think something is wrong with this exercise. The algorithm is making wrong predictions or I am making a mistake.