search

Hundredl / MG-TSD

The code for the paper "[ICLR'24]MG-TSD: Multi-Granularity Time Series Diffusion Models with Guided Learning Process"
Apache License 2.0
32 stars 1 forks source link

Hello!How can I run this model with my own dataset？ #1

Open wsp66 opened 1 month ago

wsp66 commented 1 month ago

I wanted to run the model with my own data set, but I couldn't find the code to load the local data set.

Hundredl commented 1 month ago

Thank you for your question.

Our code loads data through the GluonTS package. There are two methods you can try to load the local datasets:

  1. Integrate your dataset as a built-in dataset in GluonTS: GluonTS downloads datasets to ~/.mxnet by default. You can format your dataset to match the structure of the built-in datasets, which I believe is in JSON format, and then import it by name.

  2. Process your dataset into numpy arrays or CSV files and then load it: After importing and preprocessing, GluonTS splits the dataset into two matrices: dataset.train and dataset.test. You can directly load your data into these matrices.

Good luck!