What to take away
Divide every column by the top of a range you decided in advance, so that all of them land between 0 and 1
— this is called normalization.
Hours studied is set as 0 to 10 hours, so we divide by 10
(a person chose that range. The real maximums among the 40 are 9 hours studied and 10.5 hours slept, so nothing reaches 1).
Once the rulers match, the units disappear.
4.5 ÷ 10 and 270 ÷ 600 both give the same 0.45.
Matching up the rulers is a job for a person.
Where this pays off
When training goes nowhere, suspect things in this order: (a) the size of the inputs, (b) the step size, (c) how the data is split.