Question



Good Predictions

What makes a good prediction "good". You're predicting Y which can be either 1 or 0 - right?





Answers and follow-up questions





Answer or follow-up question 1

I understand - a good prediction is when yhat is very close to y.


Answer or follow-up question 2

Dear student,

You just answered your own question. How close y and yhat should be to be considered a good model depends on the application domain. in
stock trading a model slightly better than random is great. In other domains, such as CRM one would want very small differences between y
and yhat. And indeed, in BZAN542 we are focusing on binary responses variables, but one can estimate models for continuous and multi-class
response variables.

Best,
Michel Ballings



Sign in to be able to add an answer or mark this question as resolved.