2
$\begingroup$

I trying to predict sales quantity of particular item by using attributes of item.Not by using Time series.I have currently 3 years of past data.I am performing this by using Neural Networks.
What is the Error metric need to follow in this case ?

$\endgroup$
1
$\begingroup$

This is a fairly common task, especially in fields like economics, but the choice of using a Neural Network makes your approach different from most.

Having said that, I think you can still use common validation metrics:

  • Mean Error
  • Mean Absolute Error
  • Absolute Error ("delta")

You probably don't need anything more fancy than that, especially if those are the error metrics that your business clients will be using to measure your success (which I imagine would be the case).

Having said that, if you want a fancier ML metric to use whilst developing your algorithm you might try Root Mean Squared Error (RMSE). Here's an article on the topic.

If you wanted to, you could also construct a pseudo-R-squared metric, but those are less used with ANN's and I wouldn't choose it myself for this particular case.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.