Google Datalab: Can I query Google Cloud Datastore for trainning data for model?

0
votes

I am planning on creating a ML model using Google Datalab. I plan to keep the source data (json, structured) on datastore.

Still, I am not finding lot of examples onhow to query datastore form datalab. Is that something that can be done? Is that a good practice? Should I better write a process to send the trainning data to a CSV on Google Cloud Storage?

Thanks!

1
google-cloud-datastoregoogle-cloud-datalab
Is this question not a duplicate of stackoverflow.com/questions/34394289/… – Kolban
Thanks @Kolban. Well, in part it could be. But the second part of the question is quite different: Why there are so few examples? Isn't Datastore recommended for Datalab? Should I export it to BigQuery or Google Cloud Storage which have native support? – Gabriel
I think the reason why Datastore has fewer examples that BigQuery or Google Cloud Storage is simply that Datastore is not as frequently used as the others. That doesn't mean that it is a bad choice or that there are better choices ... it is literally the case of choosing a direction of where to go based on where you start from. If your data was on BQ or GCS to start with, then no further story. Since your data is already on Datastore, seems sensible to start from there. – Kolban
Possible duplicate of Google Datastore API from Datalab – Kolban

1 Answers

0
votes

@Kolban answered it on the comments. This is a duplicate of Google Datastore API from Datalab

Also, there are not many examples because it is not used as much as the other products

Thanks!