I’m creating dataset csv files to train LSTM network from raw data, anyone know if the first or last occurrence in the time series should be on the first or last row in the csv datafile to get the same data loading as when feeding the trained network later for predictions one by one with the last time being fed lastly into the network?
It reads the file front to back, so you should be putting your first step as the first line and your last step as the last line.