search

kaelzhang / DA-RNN-in-Tensorflow-2-and-PyTorch

A Tensorflow 2 (Keras) implementation of DA-RNN (A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction, arXiv:1704.02971)
MIT License
30 stars 12 forks source link

Recursive prediction #3

Open TDominiak opened 3 years ago

TDominiak commented 3 years ago

Hi, I wonder how I can make a forecast without using y_hist in the model. I would like to make a forecast using a recursive method. (forecasts based on the prediction of the model from the previous time step) instead of a teacher forcing.

kaelzhang commented 3 years ago

This repo is just an experiment to implement the model in the paper, but the paper did not mention recursive prediction.

TDominiak commented 3 years ago

So, I understand that if I want to make a multi step prediction, i should use single-shot approach and set accordingly ydim. At the same time, when making a forecast (in future), should I already know y-hist? This is very confusing. Thank for help.

kaelzhang commented 3 years ago

We could move discussion to #2 which is the exact same topic?

kaelzhang commented 3 years ago

At the same time, when making a forecast (in future), should I already know y-hist

That is what DA-RNN does