For the task of predicting housing prices, which activation functions could you choose for the output layer? Choose the 2 options that apply.
Why is the 2nd answer linear? Won’t that mean that housing prices can be negative?
Dear roystonlmq,
To solve that issue you could change your activation functions in the previous layers in such a way that the input to the output layer is always positive. However, a better way of thinking about this, is that your neural network will learn to output positive prices which are correlated to important features impacting housing prices.
I hope that makes thinks clear. If there are any ambiguities please let me know.
All the best,
Kiavash
3 Likes
Yeah.
The housing price will not be -ve ever since housing size can’t be less than 0 even though the linear equation can produce -ve housing price.