Evaluation using Scale transformation from old data. PaddlePaddle Framework.

[fersebas]

HI! I think there is a bug in the evaluation script or need to update

https://github.com/PaddlePaddle/PaddleSpatial/blob/main/apps/wpf_baseline_gru/kddcup22-sdwpf-evaluation/paddlepaddle/evaluation.py

Why the filename is fixed to the old one "sdwpf_baidukddcup2022_full.csv".

So when we are forecasting in this line

train_data = exp.load_train_data()

We read the old one csv and get scaler with this old data. Please check it out.

[zhoujingbo]

Hi,

Thanks for pointing out this.

The corrected file name should be "wtbdata_245days.csv". We have updated the evaluation.py file accordingly.

Thanks a lot.

Regards

[fersebas]

Thanks :)