I am currently working with a dataset with autocorrelated features, i.e, x(t) affects the value of x(t+1) which affects the value of x(t+2). Suppose this is the circuit representing x(t):

Circuit Representing X(t)

Does stacking x(t), x(t+1), x(t+2) in parallel make more sense or should they be stacked in series, or perhaps something entirely different might make more sense?

My intuition tells me that stacking them in series to have each time-step of a feature on the same wire is the best approach, but I would love to hear thoughts on this.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.