While I was exploring some tutorials (movie sentiment) I was going through some codes related to predictions, where the code was asking for DATAROBOT_KEY. However, I found that in the newer deployments, there is no mention of DATAROBOT_KEY in the Integrations tab.
Where do I find that and how the rest of the code be refactored?
Solved! Go to Solution.
Hi Santosh -
The DATAROBOT_KEY is the token that allows you to leverage your licensed prediction hardware on the cloud when submitting a scoring request to a model deployed and hosted on DataRobot SaaS. These are typically real-time scoring requests to the Prediction API endpoint on a Dedicated Prediction Engine (DPE) - our hosting servers. There are two reasons why you may not find it in some code examples you encounter.
Conversely, this article has some examples of hitting the Prediction API and the scoring resource (DPE) directly. DATAROBOT_KEY is sent over as a header item.
Thank you for your responses.
I have already checked that Predictions API tab and didn't find DATAROBOT_KEY in the code, but I could see API_KEY and API_URL only. I am sorry about mentioning 'Integrations' tab, I happened to look at Predictions API tab only.
Here, they are mentioning about DATAROBOT_KEY (looks like earlier version of DR) which I was not able to find out, so I asked here. Can I ignore that parameter? Will that still hold good for my predictions?
If not, what value should I be using for the DATAROBOT_KEY ? As you all know I am still evaluating trial version and mostly dependent on the DataRobot's servers for everything. Please let me know if I am missing anything.
Santosh, per @lhaviland 's screenshots and you being on the AI Trial (app2 host) - I think there is a 3rd case where this would be encountered. It looks like the AI Trial does not use these keys. This makes sense in that the SaaS cloud offering provides dedicated hardware for each customer to score on, however the AI Trial is a shared resource. So you can safely ignore the references to DATAROBOT_KEY.