search

zalandoresearch / pytorch-ts

PyTorch based Probabilistic Time Series forecasting framework based on GluonTS backend
MIT License
1.21k stars 191 forks source link

How can I use DeepAR in multi-variate dataset #125

Closed BEbillionaireUSD closed 1 year ago

BEbillionaireUSD commented 1 year ago

I have a customized dataset similar to the data in Multivariate-Flow-Solar (example code), which is in the form of: dataset = ListDataset(["target": np.ones(20, 100),"start": ...], one_dim_target=False), where 20 denotes the number of columns.

I am wondering how to use DeepAR to predict on my own dataset considering the DeepAREstimator doesn't have a parameter like target_dim?

Looking forward to your help. Thanks!

kashif commented 1 year ago

@BEbillionaireUSD i believe deepAR is also a univariate method, and its multivariate variant is called DeepVAR. You can use the implementation of deepVAR here or from gluonts directly which also has an implementation.

BEbillionaireUSD commented 1 year ago

Thanks very much for your prompt response! DeepVAR works.