1
$\begingroup$

Can I use the transformers for the prediction of wind power with the historical data?

Dataset

Datetime, Ambient temperature (Degree), Dewpoint (Degree), Relative Humidity\n (%), Air Pressure, Wind Direction (Degree), Wind Speed at 76.8 m (m/sec), Power Generated(kW).

15 years of data from 2007 to 2021 with a sampling time of 1 hour​

Improve this question
$\endgroup$
1
  • $\begingroup$ Hello. Welcome to AI SE. You provide some info about your dataset, but it's not clear what are your features (or independent variables) and what the label (or dependent variable, i.e what you want to predict) is. You should probably also tell us how large your dataset is in terms of observations. Finally, you should explain why you think the transformer may be a good choice for this. $\endgroup$
    – nbro
    Commented Oct 19, 2021 at 12:23

1 Answer 1

Reset to default
1
$\begingroup$

Transformers, being a general-purpose sequence model can be used for Time-Series forecasting.

There are some papers dedicated to the use of Transformer for time-series prediction and blogs.

The main ingredient for the autoregression in predictions is the mask in Transformer encoder. When the next element is predicted, tokens in the sequence attend only to the tokens back in time.

enter image description here

After each block a new element is predicted, based on the decoder and encoder tokens.

However, since the dimensionality of your data seems to be rather small, I would suggest starting from something simpler - say linear AR models or RNN, and only then work with transformers.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .