How can I efficiently identify specific columns in a large dataset using Domo’s AI/ML tools?

DataLawton

I’m working with a dataset that has 1308 columns, and I need to identify 5 specific columns that contain data like street addresses, credit card numbers, and property IDs. Or, at the very least, Domo provide 2-3 potential columns for each of the 5. The challenge here is that I'm unsure how these columns might be labeled within the data.

I’ve heard about Domo’s new AI and ML functionalities and was wondering if these tools can help me quickly sift through these columns to find the data I need. Has anyone used these features for a similar purpose? What’s the best approach to take in this situation?

ArborRose

Whew. That's a bunch of columns. Whether you do it through Domo's AI or externally….you can feed AI a list of column names and ask it the question. It may have a limit on how many you can give it at one time. You may need to feed it a hundred or so at a time and limit the answers down til you get what you need.

Let's say I go to ChatGPT or in a Domo question box….and say exactly as below:

"Working with Domo, I have a dataset with the following columns/fields:
record type,financialcategoryname,batchtimestamp,projectid,plannedcost
Can you please identify the column that best represents date, and the column that best represents amount?"

It will respond with something like this:

Certainly! Based on the column names provided:

1. Column representing date:
   • batchtimestamp: This column likely contains date and time information. It is the best candidate for representing date and time.
2. Column representing amount:
   • plannedcost: This column likely represents financial values or costs, making it the best candidate for representing an amount.

So, batchtimestamp is your date column, and plannedcost is your amount column.