How can I mark an Azure Dataset as a time series dataset reading from a parquet folder with date partitions?

Question

I would like to create a Time series dataset from a folder that contains parquet files this way:

• timestamp=2018-01-06
• timestamp=2018-01-07

How can I make Azure Dataset, through the GUI, recognises the timestamp partition as a date and mark my dataset as a time series dataset?

It is supposed to be automatic, but it doesn't work.

Answer 1

Thanks for reaching out to us.

In Azure Machine Learning Studio, you would need to setup partition format similar to python SDK, as follows, assuming your data path is "timeseries/timestamp=2020-01-01/data.parquet": Set up partition format when creating time series dataset