Github user fmcquillan99 commented on the issue:
https://github.com/apache/madlib/pull/289
```
The model table produced by the training function contains the following
columns:
gid INTEGER. Group id that uniquely identifies a set of grouping column
values.
sample_id INTEGER. The id of the bootstrap sample that this tree is a
part of.
tree BYTEA8. Trained tree model stored in binary format (not human
readable).
impurity_var_importance DOUBLE PRECISION[]. The gini impurity
importance score for the tree.
```
I don't think we need the `impurity_var_importance` for each tree in the
forest, since we have the final/averaged one on the grouping table.
And we don't put the `oob_var_importance` here so it is inconsistent.
---
