[
https://issues.apache.org/jira/browse/MADLIB-1230?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16459852#comment-16459852
]
Frank McQuillan commented on MADLIB-1230:
-----------------------------------------
When I tries with DT, I got
{code:java}
InternalError: (psycopg2.InternalError) plpy.Error: Decision tree: None of the
input features are valid
CONTEXT: Traceback (most recent call last):
PL/Python function "tree_train", line 25, in <module>
null_handling_params, verbose_mode)
PL/Python function "tree_train", line 492, in tree_train
PL/Python function "tree_train", line 271, in _get_tree_states
PL/Python function "tree_train"
[SQL: "SELECT madlib.tree_train('gcmt_cars_svec',\n
'gcmt_cars_svec_output',\n 'id',\n
'mpg',\n 'features::madlib.svec::float8[]',\n
'', -- exclude columns\n
'gini',\n NULL, -- grouping columns\n
NULL,\n 5,3,1,10\n
);"]
{code}
> DT: Database crashes when feature is an svec array
> --------------------------------------------------
>
> Key: MADLIB-1230
> URL: https://issues.apache.org/jira/browse/MADLIB-1230
> Project: Apache MADlib
> Issue Type: Task
> Components: Module: Decision Tree
> Reporter: Rahul Iyer
> Assignee: Rahul Iyer
> Priority: Major
> Attachments: dt_svec_feature_bug.sql
>
>
> In the forest_training function in the madlib schema, trying to cast a sparse
> vector into the features parameter is leading to a PANIC. Is there a way to
> allow sparse vectors to be used as part of forest training?
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
