This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
I don't have a specific solution for you, however that seems to be an error with 32 bit R. That error code is 4GB in bytes and that being the 32 bit RAM limit....
However, I have previously encountered this error I think with Forest Model and by memory @DrDan mentioned something about it being a result of R recovering from a C++ Runtime function. Even though it threw an error, did it actually complete?
This is actually not an error, it is a return code from the Microsoft Visual C++ runtime libraries. R classifies any return code other than zero as an error. When these types of "errors" are brought up on R mailing lists the answer from members of the R Core team is along the lines of "We don't know what these return codes mean, Microsoft hasn't publicly released the return code definitions, we have found that they don't impact the outcome, and they should be ignored". We have also found that they don't alter model results, and ones like the one reported here, tend to occur when machines are running low on memory, but even that is inconsistent. The one consistent return code we do see is 259, and occurs in cases where Alteryx is sending to data to R in chunks. Again, cases where that error code occurs don't produce output that is different from cases where it does not.
An R error that indicates an issue with a line of code are true errors, and they will have a return code of 1. Ones like this case, where there is no specific line of R code involved, and have a return code that is not 1 are annoying, but aren't true effors.
Is there a way to build into Alteryx at the very end that if an error code other than 1 is generated, then not to display it as an error (don't have the module state that an error has occured)? Or, maybe even better, have the module complete without error but have a warning stating something to the effect that there was an error, but it was not a true R-code error?
I am running a logistic and then stepwise regression on a total of 1.8M records where my response variable has a frequency of 25%(Yes) and 75%(No). I am also using the Oversample Field and then create samples of 50-50 for model build.
Importantly: All my variables are categorical and my total variables are about 27
Everytime i run the module (attached here), i get an error as: The R.exe exit code (259) indicated an error. And because, this is happening, i think, the Lift module is also not working and showing similar error as Lift Chart: The R.exe exit code (259) indicated an error.
Can anybody help me understand, what the exact issue is and what can we do to avoid getting this error further and build the models?
@wahmed, looking at at the screen shot of your workflow, the real problem is with the Logistic Regression tool which is erroring out upstream of the both the Stepwise and Lift Chart tools. My hunch is that the Logistic Regression tool is erroring due to missing values or other hygiene issues associated with the data. I recommend reading this blog post: http://community.alteryx.com/t5/Engine-Works-Blog/Troubleshooting-the-Predictive-Tools/ba-p/8593 by @CailinS to help you troubleshoot the error you are encountering with the Logistic Regression tool.
I wanted to follow-up on my original post. @wahmed, can you see reports in the three browse tools for the three predictive tools? Is the only stated error the R exit code of 259? If the answer is "yes" to these two questions, then there may actually be no true error at all. Instead, it may be an issue with how R itself (erroneously) responds to non-zero exit codes from the Microsoft Visual C++ runtime library, so there may in fact be not error at all.
@rlangestr, thanks for letting us know about his. We always knew that the 259 return code has something to do with memory utilization, and the fact that forcing garbage collection (via a call to gc()) to clear out memory "solves" the problem on R exit is good to know. The 259 code represents a nuisance, more than a true problem, but it is a nuisance we'd like to eliminate.