Hi folks,

Need some insights here. avoidable bias is actually unavoidable as the bias of the training error is close to the bayes optimal error or its approximation. if it is called avoidable, how to avoid it? we actually take it as the room for improvement is little compare to others.

I think I am a bit confused on how exactly the avoidable bias is defined. if it simply means the gap from training error to the optimal error. That make more sense

but per Andrew in the video " *And the term avoidable bias acknowledges that there’s some bias or some minimum level of error that you just cannot get below which is that if Bayes error is 7.5%, you don’t actually want to get below that level of erro*r", it seems like avoidable bias here means the gap between bayes optimal error and the best the model can get.

Thanks for your thoughts!