Re: [Gretl-devel] Mean Absolute Percentage Error: possible mistake
On Fri, 21 Oct 2016, Henrique Andrade wrote:
Dear Allin and Jack,
I think there is an error in the definition of the Mean Absolute Percentage
Error (chapter 29, section 29.4). I think we should use the modulus of the
entire division:
abs(e_t/y_t)
Now we have:
abs(e_t)/y_t
Thanks, you're right. That typo is now fixed in git. (The actual
calculation was right, it's just the doc that was wrong.)
Allin