Fortran run-time error when using statsmodels in Python

Question

I am attempting to generate a fairly simple ARMA model using STATSMODELS in Python (Pyscripter IDE). I kept getting an odd error, so I took a step back and tried the following example. I get the same error.

http://statsmodels.sourceforge.net/devel/examples/notebooks/generated/tsa_arma.html

Any information would be helpful. Full disclosure, I'm a scientist but not a COMPUTER scientist.

I get the following error when running the script:

Intel(r) Visual Fortran run-time error

forrtl: severe (9): permission to access file denied, unit 8, file C:\Program Files (x86)\PyScipter\iterate.dat

Image              PC          Routine    Line       Source
libifcoremd.dll    07DA1A70    Unknown    Unknown    Unknown

...
_lbfgsb.ypd        08344077    Unknown    Unknown    Unknown
...

Answer 1

my guess

iterate.dat is the logfile for the optimization that scipy fmin_lbfgsb writes. And the fortran extension that implements lbfgsb doesn't have permission to write to C:\Program Files

possible solutions, I'm not sure which will work:

• run the script in a local working directory where you have write access.
• suppress the logfile, iprint=0 as fit option might work http://docs.scipy.org/doc/scipy-0.15.1/reference/generated/scipy.optimize.fmin_l_bfgs_b.html (maybe disp=0 works also)
• use a different optimizer that doesn't print from Fortran (only as last resort, e.g. with bfgs )

Answer 2

I faced the same issue today and the issue was solved by doing two different things I found on StackOverflow.

model.fit(iprint=0, disp=0)

Use the following two parameters:

1. iprint=0
2. disp=0