i have a question about multi step forecasting result. for example, i have a dataset from 1 to 100. I applied sliding window technique and i used 24 timesteps to predict 20 future data. i had following samples:

sample 1 input: (1 to 24 (inclusive)) output:(25 to 45) sample 2 input: (2 to 25 (inclusive)) output:(26 to 46) sample 3 input: (3 to 26 (inclusive)) output:(27 to 47) .... sample n

When I fit my model and make a prediction, there will be a conflict with output data because it will produce a sequence of data that contains data 27 from sample 1, another from the second sample, and the last from the third sample. End of the forecasting, I will have 3 data for the 27th data; which one should I consider as forecasted data?

thank you



Your Answer

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