The Graph node only works on a sample of data. Sometimes (depending on data size) I would need to produce graphs on all the data.
Perhaps this restriction needs to be removed. There is still a cache option in properties (for the graph node) to base it on a sample.
I still get asked this even if I have generated a sample on the Data Source node (an option in properties).
So this seems to be duplicating the work. In the case where the Data Source node is run with sampling and then the Graph node does sampleing. Is the Graph node sample created from the Data Source sample (i.e. a sample of a sample) or is it sampling the original data (it does the sampling for a second time)?
A few more suggestions regarding the graphs, including
- to make the graphs moveable, so that we can re-position the graphs in the different orders
- can have different numbers of graphs on each line. Allows the grouping of related graphs together
- can scale the graphs to a suitable size (i.e. to make them slightly bigger if required). This is in addition to the current full zoom.
The ODM help in SQL Dev says the following for the SQL Node
Below the text area are:
Columns, list the columns and data types
Preview, displays a small number of the rows returned by the query
This area and preview does not exist on my SQL node
There does seem to be something below the query text area as I get the mouse cursor to change to indicate that I can change the size of the area. But nothing happens when I try this
There is a technical requirement that requires a table or view to be constructed as the source the graph node.
So currently the cache option is creating a sampled table.
What ever down stream sql etc. that was generated would proceed this sample.
So if you have a Sample node prior to the Graph node that creates say a sample of 100,000 rows.
The Graph node would then sample that down to a smaller set.
We are now looking at an option where the user could check an option on the Graph node to use all the data.
In this case, at View would be created to resolve the technical requirement, and the user would have access to all the data from the parent node.
I think this addresses your request.
Check out the SQL Dev OTN forum. It seems there are some performance issues noted in the SQL DEV 4.0 EA1 release.
This should be addressed as part of SQL Dev 4.0 EA2 or RTM release.
You should also consider either upgrading to 12.1 db or 126.96.36.199 (which should be out soon) as they will include significant improvements in the Data Miner server implementation.
See the following link for some server documentation on Dynamic Scoring (aka Predictive Queries):
I've opened SQL Dev this morning and it works. I then opened a ODM connection and Workflow. The various containers/tab for ODM open in positions that were not previously used and then SQL Developer hangs. I cannot do anything in SQL Dev/ODM. I cannot click on anything, move anything. It does not respond. I have to use task manager to kill SQL Dev
Can you think of any changes made to your db or client that may have affected this change in behavior?
Can you do the following:
1) Launch SQL Dev with console:
3) View->Extension Diagnostics
4) Proceed to the Data Miner tab, open connection and workflow and monitor Log, Extension Diagnostics and Console for any relevant messages and post results here.
5) Also go to the bottom the db alert log and see if there are any db query or system failures as well, that occurred at the same time.
After that, you can also try rebooting client and bounding db to see how that affects behavior.
I have made no changes to the 12.1 DB. I had closed SQL Dev cleanly, after adding a model details and create table nodes. These ran OK and produced the output.
I had shut down the DB and SQL Dev
It was after starting the DB and SQL Dev that the error occurs.
If I leave SQL Dev/ODM for approx 5 minutes it becomes usable.
The window title area, where the SQL Dev icon is, keeps flickering with two different messages, which unfortunately I cannot read.
I will try your suggestion for getting the log