What are the criteria which can be used to evaluate the performance of online machine learning methods in comparison with the batch learning methods(classification)?
Prediction statistics like accuracy, precision etc serve as criteria. Also, time taken & RAM-hours required for training can also serve as criteria. Regarding RAM-hours, how is it possible to measure RAM hours when the execution happens in a windows machine in R environment?
Is there any other criteria which can be used to compare online and batch machine learning methods?
