That's a great point, Oleksandr. However, note that we are using BOTH sine and cosine feature. While sin(2AM) = sin(10AM), cos(2AM) ≠ cos(10AM). Thus, when one of these features falls into the problem you mentioned, the other is there to make the distinction.
This so neat and clever! Thanks a lot
But this introduces another problem - each of the, let's say, hours have a counterpart with the same feature value.
For example, 2 AM is the same as 10AM for the model. Are there any workaround for this as well?
That's a great point, Oleksandr. However, note that we are using BOTH sine and cosine feature. While sin(2AM) = sin(10AM), cos(2AM) ≠ cos(10AM). Thus, when one of these features falls into the problem you mentioned, the other is there to make the distinction.