I was doing machine learning course at coursera and there was a lecture about diagnosing high bias, or high variance from learning curves. If anyone interested here is the lecture - https://class.coursera.org/ml-003/lecture/64. I just can't find it enywhere else, I mean in some serious literature. So my question is, if this is really good, or commonly used approach to diagnose bias - variance in classification task? Or are there any commonly used, probably preferable way how to determine bias, variance error part in classification task?
Any comment, point to useful resource would be very appreciated.