zhycs / last Goto Github PK

I hope this message finds you well. First of all, I would like to express my appreciation for your remarkable work presented in your paper titled [Learning Latent Seasonal-Trend Representations for Time Series Forecasting]. The findings and methodologies described in the paper have left a lasting impression on me.

However, I am currently facing some difficulties in reproducing the results reported in your paper, specifically regarding the predictions made on the ETTh1 dataset. According to the paper, when the prediction length is set to 24, the Mean Squared Error (MSE) is reported as 0.324. However, when I attempted to replicate these results using the provided source code, I obtained an MSE of 0.3408.

Considering the variance in the results, I would greatly appreciate it if you could provide some insights or guidance on adjusting the parameters for different datasets. Specifically, I am interested in knowing which parameters should be adjusted to improve the reproducibility of the results on different datasets, such as ETTh1.

Thank you very much for your time and attention to this matter. I look forward to your kind response and any assistance you can provide.

The target variable used by CoST and TS2Vec on the electricity is "MT_001", while the target variable used by LaST is "MT_369", Could you check the experiment results? Thanks.