Log to Data, Missing Attribute Names
Hi,
After converting log values to an example set with "Log to Data" operator, following operators (such as aggregate) cannot catch the attribute names of the set. In the meta data information correct number of attributes are counted but their names are not shown and hence unusable. Is there a solution for this?
However, if I type the attribute names correctly in operators like "Generate Attributes" the process runs without error. But I cannot enter or type any name in the "aggregate" operator, only selection is allowed.
Best Answer
-
Thomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,761 Unicorn
Ok, in some situations the metadata doesn't properly propogate through the process. It happens from time to time.
To solve this, you should toggle on the "Syncronise MetaData with Real Data" option in Process pull down menu. Run the process once and it back fills the meta data and you can then see your attribute names populate.
You can also try the Materialize Data operator, that drops the data out of the background view. This sometimes helps, but in your process it did not (just ignore them in your process below).
2
Answers
If you could post and XML of the process, we could troubleshoot your trouble.
Hi Thomas,
My real process is rather big. So, I attach a similar process. The problem is the same. As you can see, I can use log column titles as attribute names in "Generate Attributes" operator despite being not able to select any of them. Yet, In "Aggregate" operator there seems to be no solution to group because it compells me to select existing attributes which do not appear.
Thank you Thomas for your help,
Your suggestion did really help in my example. As I said before, my real process is complicated. Synchronization did not help at first, but I solved it afterwards. The log and log to data operators were in testing part of a validation operator. I sense(!) that the problem might be that log to data is inside validation, so I took it out from there and just remember the log, outside the validation I recalled the log converted it to data. It worked!
Thanks again. But I am curious about one thing now. Why is data synchronization not checked default? Is there any drawback of doing so?
It's not checked on by default because 99% of the time the meta data propogates. Earlier versions of RM Studio did not have that feature and it was added in later to help solve the 1% of the problems, which can be rather annoying.
I am facing the same issue as mentioned above by @suleymansahal. And also tried both the solutions given by @Thomas_Ott. but It's not working.
Awaits other solutions(if any).
Thank-you