I’ve been very interested in Microsoft’s Machine Learning offering that they’ve recently launched on Azure as it looks like it will bring data-mining out of the stranglehold of data scientists and much closer to data users, so I eagerly set about viewing the online tutorials and webcasts Microsoft have provided.
Unfortunately, this particular webcast rather glossed over the step where you access the browser based front end interface, called ML Studio by switching to a browser tab where Machine Learning (ML) Studio was already open. It soon became clear why.
When I tried to launch ML Studio directly from the Azure Portal wherein I had created my Machine Learning workspace I was prompted to log in.
Naturally, I used the account I had assigned as the creator of the container. Nothing doing; ‘That Microsoft account doesn’t exist. Enter a different email address or get a new account.’ as figure 1 shows:
Strange. The account absolutely does exist otherwise I would not have been able to access the Azure Portal to create the ML container in the first place!
After a bit of to-and-fro on the web (and some blind alleys involving cookie deletions etc) I eventually found this Machine Learning forum posting. Yeah. ML Studio does not yet support Organisational Accounts. The very account you need to set up Azure and the ML container to use in ML studio.
I had to go back into the Azure Portal, create a new ML container (as the owner of an existing container cannot be changed as of the time of writing this post) and set the owner of the container to my private Live account email, and then I was able to log into ML studio.
If this was mentioned in the tutorials/webcasts it would have saved me over an hour of wasted time.
Hopefully this will save you a little time, at least until MS enable organisational account support.