I need to download a custom dataset in an Azure Jupyter/iPython Notebook. My ultimate goal is to install an R package. To be able to do this the package (the dataset) needs to be downloaded in code. I followed the steps outlined by Andrie de Vries in the comments section of this post: Jupyter Notebooks with R in Azure ML Studio.
Uploading the package as a ZIP file was without problems, but when I run the code in my notebook I get an error:
Error in curl(x$DownloadLocation, handle = h, open = conn): Failure when receiving data from the peer Traceback:
- download.datasets(ws, "plotly_3.6.0.tar.gz.zip")
- lapply(1:nrow(datasets), function(j) get_dataset(datasets[j, . ], ...))
- FUN(1L[[1L]], ...)
- get_dataset(datasets[j, ], ...)
- curl(x$DownloadLocation, handle = h, open = conn)
So I simplified my code into:
library("AzureML")
ws <- workspace()
ds <- datasets(ws)
ds$Name
data <- download.datasets(ws, "plotly_3.6.0.tar.gz.zip")
head(data)
Where "plotly_3.6.0.tar.gz.zip" is the name of my dataset of data type "Zip". Unfortunately this results in the same error. To rule out data type issues I also tried to download another dataset of mine which is of data type "Dataset". Also the same error.
Now I change the dataset I want to download to one of the sample datasets of AzureML Studio. "text.preprocessing.zip" is of datatype Zip
data <- download.datasets(ws, "text.preprocessing.zip")
"Flight Delays Data" is of datatype GenericCSV
data <- download.datasets(ws, "Flight Delays Data")
Both of the sample datasets can be downloaded without problems.
So why can't I download my own saved dataset?
I could not find anything helpful in the documentation of the download.datasets function. Not on rdocumentation.org, nor on cran.r-project.org (page 17-18).
It seems the error I got was due to a bug in the (then early) Azure ML Studio.
I tried again after the reply of Daniel Prager only to find out my code works as expected without any changes. Adding the id and auth parameters was not needed.