I have to make prediction on streaming data for which I have trained the model offline.
Is it a common practice to use scaling when we have to train the model offline and have to predict the on streaming data, also how to scale the streaming data?
I have to make prediction on streaming data for which I have trained the model offline.
Is it a common practice to use scaling when we have to train the model offline and have to predict the on streaming data, also how to scale the streaming data?
Scaling of streaming data must be based on your training set parameters. Your model has learnt based on that scaling, therefore, in order to assess its generalization power, you have to repeat the same scaling that your model learned from.
you can find similar case in Machine Learning with train and test data set here