What is the purpose of defining a y_range(as mentioned in chapter 6)?
What would happen if we did not define a y_range?
Hi Tom
Classification returns a single or multiple categories.
Regression returns a value using sigmoid (-1,1).
y range allows you to adjust the sigmoid as sigmoid * (hi-lo) + lo. I think Jeremy mentions somewhere rather than -5,+5 to use -5 to 5.5 because sigmoid only reaches +1 .
So if you are predicting crop yield based on sunshine, rain, fertiser, soil type, pests,… you need a value other than (-1,1) but it is an interest questions you would choose the range.
Regards Conwyn