I have been looking at how to split my data for training/validation/test for a timeseries using LSTM and have some conflicting thoughts I would like to get a bit more clarity on. I came across: QA1 and QA2
- Firstly isn't walk-forward somewhat redundant for Long Short Term Memory Networks?
- Initially I was under the impression that I should shuffle my input sequences, but wouldn't I achieve the same effect of "walking" if I hadn't shuffled? (Or am I wrong to shuffle in any scenario?)