Keep Columns After Using Group By Tile in ETL

deona720

I haven't played around with the group by tile yet so this might be a very simple answer I overlooked. But why do all of my columns disappear and leave only the columns I identified in the group by tile from my data after I use this function? I know I could join the data back in but I was assuming I was using it wrong. This is my etl -

ColemenWilson

Because they have now been aggregated and Domo doesn't know how to aggregate the remaining columns. So if you include them in the group by then they aggregate as much as they can with those fields, example:
- You have dates and a category included in the group by and individual transactions summed. Domo will aggregate (sum) to the highest level of unique combination of date and category. As you add more fields, it changes the grouping. Examples illustrating this below:

Raw Data:

Group by 1: Date

Category can't be added here because both categories exist for both dates, so to group by date means excluding category.

Group by 2: date and Category

GrantSmith

Depending on your groupings and the metrics you're calculating you can either group by and then join it back to your original dataset like you mentioned or possibly use a Rank and Window tile to do your aggregation within your entire dataset and leave your columns in place. But @ColemenWilson mentioned, aggregation drops it because it doesn't know what to do with all your other columns.

ColemenWilson

Because they have now been aggregated and Domo doesn't know how to aggregate the remaining columns. So if you include them in the group by then they aggregate as much as they can with those fields, example:
- You have dates and a category included in the group by and individual transactions summed. Domo will aggregate (sum) to the highest level of unique combination of date and category. As you add more fields, it changes the grouping. Examples illustrating this below:

Raw Data:

Group by 1: Date

Category can't be added here because both categories exist for both dates, so to group by date means excluding category.

Group by 2: date and Category

GrantSmith

Depending on your groupings and the metrics you're calculating you can either group by and then join it back to your original dataset like you mentioned or possibly use a Rank and Window tile to do your aggregation within your entire dataset and leave your columns in place. But @ColemenWilson mentioned, aggregation drops it because it doesn't know what to do with all your other columns.

DataMaven

@deona720 - Feedback on the Dataflow - Is there a reason you are using so many append tiles? I strongly recommend appending all of them in one append tile. This will not only be more efficient, but will allow you to easily review them all at once for any field conflicts.

Additional recommendation - Name each of those formula tiles so that you can tell which is which in your append.