Solar radiation prediction using LSTM

Created an LSTM model for day ahead forecast of solar radiation

Here accuracy seems to be low but their RMSE MAE R2 are good. Is this correct?


Hi @sowkarthika_s,
Accuracy can vary tast to task, and data to data.
If loss is minimizing, this is indeed a good result.
You can compare these results with several other existing Timeseries models for a better comparison.

1 Like

Thank You Sir. But I wanted to know whether these model is acceptable or not? Hyper parameters are tuned using grid search method

1 Like

Hi @sowkarthika_s,
Yes, that’s what i meant to say… we can only select or reject some model based on a comparison of results by other methods/ models.

Good to know you are using your DS skill in GHI/Solar Power prediction. You can check the accurracy in terms of time-blocks (say +/- 15 % accuracy band). You have options to use dynamic weather model forecast data in real-time prediction.