The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
Join on 2 Joins produces Error
Hi,
each time I try to join examplesets which are the output of joins I get an "Error transforming meta data transformation: java.lang.NullPointerException". The process seems to run correctly, but while designing the process it's kind of irritating. Is this behaviour a result of a mistake I do, or is it a Bug?
Forgot to mention:I use version 5!
Greeting from snowy Lippe/Germany
Peter
each time I try to join examplesets which are the output of joins I get an "Error transforming meta data transformation: java.lang.NullPointerException". The process seems to run correctly, but while designing the process it's kind of irritating. Is this behaviour a result of a mistake I do, or is it a Bug?
Forgot to mention:I use version 5!
Greeting from snowy Lippe/Germany
Peter
0
Answers
this seems to be a bug in the meta data transformations. Would you be so kind to post a example process? I will then take a look at that.
Greetings,
Sebastian
what a fast reply!
I do not understand this XML-stuff but I think it's the better way to post the process. It does nothing but joining examplesets generated by queries on Microsoft Acess tables. The last join delivers the data I expect but the metadata is not generated: Ouch, I definitly like the Process-view more than this.
Hope you can read more than I do out of this code.
Thanks for your patience with noobs like me, for your great software (which I can't handle yet) and for the fantastic Webinar you held in Dec last year.
Best regards
Peter
if you think, I'm answering quickly, try our enterprise support, it'll show you, what's really quick
I have loaded your process and exchanged your access reader by data generators (Since I didn't have your files). Now it looks like below and works just fine. Please try to update to the newest RapidMiner version 5.0.003. If the error occurs again, please tell me. Greetings,
Sebastian
I'm just trying out RapidMiner at home but if I manage to get good results on my data - which is real data from my job - it might end in business-relationship to Rapid-I and then I will test how quick quick can be
I updated to the newest version - but nothing changed.
So I shrinked my data from >=1000 examples (Ifear even more than 10000) in each exampleset (outputs of joins too) to something between 24 and 264 (some this amount, others that).
Now my process dosn't through any error! ;D
It seems that metadata is not generated if there is too much data.
Maybe you can verify this guess?
Thanks in advance
Peter
the number of examples should be completely irrelevant to the meta data transformation. Where does the error occur? Is it shown in the Problems tab or does it occur during process execution?
Unfortunately I cannot reproduce the problem here with the generate sets from the process above. Do you have nominal values in your data set?
Greetings,
Sebastian
The error is shown in the Problems tab - during process execution there is no error (and results seem to be ok)!
My data set consists of nominal and numerical values, some attributes with missing values (i.e. null).
If the number of examples is irrelevant, it must be the values of specific examples. I will try to identify these examples next weekend and will inform you afterwards.
Greetings
Peter
Anyone joining 2 data sets with differently named ID columns?
-Gagi
this is indeed annoying. I do my best to find this bug, but I still cannot reproduce it. You are all using the final version 5.0.003, I guess?
This process does work on my side, does it make problems if you load it? Greetings,
Sebastian
yes now I use 5.0.003 but nothing changed. So tested a bit.
I think, I found some reasons for this behaviour in metadata:
1. In Joins the amount of examples in the resulting exampleset is determined by the right part as far as Metadata is concerned.
2. Read from ODBC or Excel only shows the exact number of example, while it's less than 1000 (otherwise metadata says >=1000 examples).
3. Metadata with the information >=1000 examples seems to be faulty.
4. Changing The join from inner to left or right changes nothing in metadata.
To prove this, I modyfied Sebastian's process: My Excel inputfile is just filled in the first column with numbers 1 to 1000 (First row is name of Attribute= Lfdnr).
Btw., is there a chance to get read fom odf-files (Open office)?
I hope, now you can reproduce the bug.
Greetings
Peter
I finally found it! Thanks a lot. The bug does not occur anyway, but the meta data is deleted anyway. Don't know if this is the best one could do there, we will check that.
Greetings,
Sebastian
Two example data sets (differing attributes, same ID column) connected both (multiply) with i) a Set Minus Operator and ii) an inner join, finally connected to an union operator? It also throws a null pointer when "Set Minus" is empty. This is when the first example set completely matches the second.However, it does not, if both sets differ by one entry. Is the union operator intolerable to this case, is it this bug or didn't I find the right operator?
Cheers - Oliver
this seems to be a bug. If you would design a small example process using Generate Data operators as data sources and post it here inside this #-button's code area? It would make my life easier fixing this issue.
Greetings,
Sebastian
Is this helpful to you?
Greetings,
Oliver
http://www.mediafire.com/file/mwiiknkuv14/DataGenA.xls
http://www.mediafire.com/file/umozx3omjqn/DataGenB.xls
the process with your data does not cause any error at my RapidMiner, so I guess the bug has been solved.
Greetings,
Sebastian
I was too fast, I guess you did not distribute the new version yet :-) I still got the null.pointer exception. I will test again when you distribute the new version. Keep up the good work!
Greetings Oliver