I agree, having a data model defined and documented would help a lot in separating processing from a specific ingest flow.
Michael On Thu, Jun 22, 2017 at 1:31 PM, Jonathan Natkins <[email protected]> wrote: > Personally, I'd love for there to be more information about the expected > schema for the ML jobs, as well as information about where the data can be > picked up from. The documentation seems to be mostly written with a > specific example in mind, so is not extremely helpful when trying to > integrate new data sources. A data dictionary would help with being able to > map fields from data formats (other logs, etc) to fields that spot-ml can > process. > > Whatever happened to the open data model that was being discussed for Spot? > > Thanks! > Natty > > On Thu, Jun 22, 2017 at 10:10 AM Barona, Ricardo <[email protected] > > > wrote: > > > Hi everyone. > > > > I’m happy to see how more people is playing with Spot and particularly > > with spot-ml everytime. > > > > Something that I’ve noticed thanks to these two Jira issues ( > > https://issues.apache.org/jira/browse/SPOT-149 and > > https://issues.apache.org/jira/browse/SPOT-174) is that sometimes users > > are going to want to try spot-ml without ingesting data using spot-ingest > > and I think that’s cool but seems like that can lead to inconsistent > schema > > issues. > > > > I’d like to know what you think, what would be the best approach to deal > > with this; I’m thinking that we can add schema validation to spot-ml > before > > anything else happens but I don’t know if that’s going to lock things too > > much. > > > > Please share your thoughts. > > > > Thanks, > > Ricardo Barona > > > -- > Jonathan "Natty" Natkins > StreamSets | Field Engineering Director > mobile: 609.577.1600 | linkedin <http://www.linkedin.com/in/nattyice> > -- Michael Ridley <[email protected]> office: (650) 352-1337 mobile: (571) 438-2420 Senior Solutions Architect Cloudera, Inc.
