What does HackerNews think of reckon?

Reckon works well enough for me.

https://github.com/cantino/reckon

Thanks for the post Sagar, I'm the maintainer of reckon [1], a tool to help categorize transactions, which I used in ledger. There are a lot of interesting tools for doing plain text accounting [2] and I'm always interested in learning about new ones.

Reckon uses TF/IDF with cosine similarity, but I would be interested to see how you use Random Forest. Please post your code somewhere, I'd love to see it and learn something new!

[1] https://github.com/cantino/reckon [2] https://plaintextaccounting.org/

I handle transfers between accounts manually - as in when I see duplicates show up, I remove one of them. A neat thing that ledger-reconciler does is to print the balance (as listed on the banking website) for each account. That information combined with Ledger's balance report tells me immediately whether something is off and if investigation is needed.

This in itself has saved me so much time that I can deal with manually editing the occassional duplicate entry (internal account transfers). I also keep all my ledger data in git so it's very easy to see what has changed, etc.

Have you looked at the reckon gem[0] for auto-classification? That is basically what this program uses.

[0]: https://github.com/cantino/reckon

Good on you for persisting with it.

I developed a mobile manual entry personal finance tracking application as well, stuck with it for two years, but I'm back to manual entry using Ledger because of the pain in the ass it is to enter data on mobile, whereas I can use automation to import my various bank accounts data (e.g. https://github.com/cantino/reckon) with Ledger.

The reporting on the data is the gold, for me, the mobile app didn't really add that much value, so I just gave up.

https://github.com/cantino/reckon imports CSV to Ledger, with Bayesian automatic categorization.

You might let people download their bank's CSV dumps and parse them with something like reckon-- https://github.com/cantino/reckon

> - I found it super laborious to enter transactions. I guess I grew up post-checkbook but when I'm not near my computer for a few days and have a pocketful (or not) of transactions to enter, I would get behind and then just give up. Even with logging into my bank account to cheat.

I had the exact same gripes with ledger. Although there are means to pull csv-data from most banks, one still would have to categorize the entries on each import into ledger.

There is https://github.com/cantino/reckon which claims to solve this by automatically sorting entries into the correct accounts by means of Bayesian learning. \nSadly the machine learning worked unreliably with my data - I would still have to manually sort many of the recurring and previously "tagged" entries into the proper account each time they came up in the csv-file.