Matthias Boehm created SYSTEMML-1824:
----------------------------------------
Summary: Wrong transformapply output with col name specification
and subset of cols
Key: SYSTEMML-1824
URL: https://issues.apache.org/jira/browse/SYSTEMML-1824
Project: SystemML
Issue Type: Bug
Reporter: Matthias Boehm
Given a frame input with column names {{name1, name2}}, the following script
produces incorrect results:
{code}
X = read("X.csv", data_type="frame", format="csv", header=TRUE);
spec = "{ids: false, recode: [ name1, name2 ]}";
[Y,M] = transformencode(target=X, spec=spec);
spec2 = "{ids: false, recode: [ name2 ]}";
Z = transformapply(target=X[,2], spec=spec2, meta=M)
print(toString(Z));
{code}
The output is supposed to be 1x3 and properly recoded but currently returns
{code}
NaN
NaN
NaN
{code}
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
