Inconsistent value in reading xpt

[guoranba]

Hi all,

I run the below 2 codes to read datasets in xpt format and got different values for TRT02PN. It should be NaN for SUBJID 17-004 but it’s 9.380616e-312 when reading ds1.xpt.

Code 1:

ds1tup = prs.read_xport("ds1.xpt",encoding="euc-cn")  
ds1=ds1tup[0].copy()
ds1=ds1.loc[ds1.SUBJID=="17-004"].copy()
ds1.head()

Code 2:

ds2tup = prs.read_xport("ds2.xpt",encoding="euc-cn")  
ds2=ds2tup[0].copy()
ds2=ds2.loc[ds2.SUBJID=="17-004"].copy()
ds2.head()

Both ds1.xpt and ds2.xpt have 3 variables: SUBJID, ASEQ, and TRT02PN. There are 127116 rows in total for ds1.xpt while, for ds2.xpt, it's the subset of ds1 and there are 1334 rows which I only kept the observations with SUBJID in ("01-001","01-076","16-044","17-004").

Does anyone have any ideas of what’s happening? Thanks in advance.
ds1.zip
ds2.zip

[ofajardo]

It is a bug, I created the issue #177

[guoranba]

Thanks ofajardo. Looking forward to the next release.

[ofajardo]

Thanks for the very reproducible and nicely written report!

[ofajardo]

it should be now fixed in the new release 1.1.5

[guoranba]

It works now. Really appreciated!