--* LDA_arrayify_input:
--*   attr:
--*     fillcolor: '3'
--*   desc: Shapes data into arrays of concept_ids for model training.
--*   ext: sql
--*   inputs:
--*   - INPUT_persons_datapartners_concepts
--* 

-- we're going to need the concepts for each person_id
-- in a single array-column entry

SELECT person_id, data_partner_id, collect_list(string(concept_id)) as concept_ids, hold_out
FROM INPUT_persons_datapartners_concepts
GROUP BY person_id, data_partner_id, hold_out