Update `ingest` stage to use `noctua` `unload = TRUE` option about model-res-avm HOT 3 OPEN

@dfsnow - Is there a particular output that you want for the value / treatment tests? I ran the ingest script twice, once with the unload = TRUE and once without the unload = TRUE. I then sorted the columns (they were ordered differently which is something to note), and used dataCompareR to test the different outputs. The only differences were related to small decimal disparities.

from model-res-avm.

@dfsnow - Is there a particular output that you want for the value / treatment tests? I ran the ingest script twice, once with the unload = TRUE and once without the unload = TRUE. I then sorted the columns (they were ordered differently which is something to note), and used dataCompareR to test the different outputs. The only differences were related to small decimal disparities.

@Damonamajor As mentioned in the issue, I would specifically focus on the categorical columns that are stored in Athena as arrays. That's mostly anything with a loc_tax_ prefix. I'm worried that the string processing we set up for the previous method may interfere with the new one and give us different results.

from model-res-avm.

@dfsnow - Is there a particular output that you want for the value / treatment tests? I ran the ingest script twice, once with the unload = TRUE and once without the unload = TRUE. I then sorted the columns (they were ordered differently which is something to note), and used dataCompareR to test the different outputs. The only differences were related to small decimal disparities.

@Damonamajor As mentioned in the issue, I would specifically focus on the categorical columns that are stored in Athena as arrays. That's mostly anything with a loc_tax_ prefix. I'm worried that the string processing we set up for the previous method may interfere with the new one and give us different results.

Those columns didn't ping anything, but I'll take a deeper dive into them.

from model-res-avm.