Github user kaknikhil commented on a diff in the pull request:
https://github.com/apache/madlib/pull/250#discussion_r177928648
--- Diff: src/ports/postgres/modules/convex/mlp_igd.py_in ---
@@ -667,7 +678,8 @@ def _validate_dependent_var(source_table,
dependent_varname,
if is_classification:
# Currently, classification doesn't accept an
# array for dep type in IGD
- _assert("[]" not in expr_type and expr_type in
classification_types,
+ _assert(("[]" in expr_type and expr_type[:-2] in int_types) \
--- End diff --
This check does not differentiate between a 1d array and a 2d array. If you
pass a 2d array as the dep variable, both train and predict functions go
through and here is what the output of predict looks like
```sql
madlib=# select * from mlp_pred_dep;
id | estimated_prob
-----+----------------
2 | {1}
13 | {1}
15 | {1}
28 | {1}
33 | {1}
```
We don't have to necessarily worry about this, we can just add a note in
the user docs but I just wanted to mention it. What do you think ?
---
