I believe I’ve send the end of Autocat and it’s called Numerous AI. Using its NUM.INFER function I can seed the AI with past categorization data (or any other field I want) and the AI does its work. I can build a huge database of client and personal descriptions and categories and have it search that for newly connected sheets.
Questions -
Anyone else trying this with success or not?
Is there still an issue with formulas in Excel transaction sheets?
Sounds interesting, @psavini. Other users have begun to share concepts around AI and ML for categorization (like this one)… Let us know how it goes.
Let me just add some context, I didn’t mean to knock AutoCat, but for me it’s never really worked 100% the way I want. I have multiple columns/attributes that cascade, so when my electric bill goes in as category Utilities (it can do this) the next pass should be classifying Utilities as Home for the group attribute. This last part it has always seemed to struggle with.
With Numerous I was able to write a simple formula that seems to work -
=NUM.AI(“Classify the transaction description “&””“”&[@Description]&“”“”&" into one of the following categories and only return the category: "&ARRAYTOTEXT(category_list))
category_list is a named range of categories
You can also use the NUM.INFER to shorten descriptions so things like “Aplpay Trader Joe S” will shorten to “Trader Joe S” and so on. At one time I used the Short Description column but there were too many errors in there for it to be reliable.
No offense taken. I fully endorse your experimentation. We’d love to make Autocat more nimble and customizable.