The Pros and Cons of Deep Learning Models for Time Series for Casting

A simpler model is easier to implement, less prone to have bugs in it. Sometimes a gipart take also quite memory size and computational power as well. Maybe a competitionally, sometimes deep learning models can be competitionally more efficient. But we don't want to argue in our paper that deep learning is senseless or doesn't make sense in time serious for casting. We want to challenge the way deep learning models aren’t and may be also challenge further or upcoming deep learning models for time series for casting should just perform a more rigorous, experimental set up.