How to Train a Predictor Using a Time Series?

The technique that we are using. Usually we work with like, it depends on the window length, primarily. We don't have any hard restriction for it to work. So as long as you have sufficiently enough, not very large, like some thousand a points, or that would be enough for like a 60 look back window to get a prediction for like ten points ahead. With this technique that we is using. You create a tenser from these windows, like you have many windows, and you create batches, and then you create your tenser, and you train it. And then, based on your window length, you can make the next prediction.