Smart Categorize

Smart Categorize

Overview

Automatically categorize uncategorized transactions by matching vendor/description patterns against known rules. Uses existing categorization rules first, then suggests new rules for unmatched transactions based on common vendor names.

Wilson Tools Used

• categorize — apply pattern-based categorization rules to transactions
• transaction_search — find uncategorized transactions

Workflow

1. Use transaction_search to find all transactions where category is null or empty.
2. Run categorize to apply existing categorization rules (pattern matching on description field).
3. Report how many transactions were categorized by existing rules.
4. For remaining uncategorized transactions, group by vendor/description similarity.
5. Suggest category assignments for each vendor group (e.g., "SPOTIFY" -> Entertainment, "SHELL OIL" -> Transportation).
6. Ask the user to confirm or adjust the suggested categories.
7. Apply confirmed categories and optionally save new categorization rules for future imports.

Without Wilson

You can categorize transactions manually in a spreadsheet:

Setting Up Category Rules in Excel/Sheets

1. Create a reference sheet called "Rules" with two columns: Pattern and Category.
2. Add your vendor patterns:

| Pattern | Category | |---------|----------| | AMAZON | Shopping | | WHOLE FOODS | Groceries | | SHELL | Transportation | | NETFLIX | Entertainment | | STARBUCKS | Dining |

1. In your transactions sheet, use a lookup formula in the Category column:

• Excel: =IFERROR(INDEX(Rules!B:B,MATCH("*"&"AMAZON"&"*",Rules!A:A,0)),"Uncategorized") — but this only works for exact matches.
• Better approach with Excel: Use a helper column with =SUMPRODUCT or VBA macro to do partial matching.
• Google Sheets: =IFERROR(VLOOKUP("*"&A2&"*",Rules!A:B,2,FALSE),"Uncategorized") does not support wildcards in VLOOKUP.
• Practical Google Sheets approach:

`` =IF(REGEXMATCH(A2,"(?i)amazon"),"Shopping", IF(REGEXMATCH(A2,"(?i)whole foods|trader joe"),"Groceries", IF(REGEXMATCH(A2,"(?i)shell|chevron|exxon"),"Transportation", "Uncategorized"))) ``

Common Category Mapping

| Vendor Pattern | Suggested Category | |---|---| | AMAZON, TARGET, WALMART | Shopping | | WHOLE FOODS, TRADER JOE, KROGER, SAFEWAY | Groceries | | UBER EATS, DOORDASH, GRUBHUB | Dining | | NETFLIX, SPOTIFY, HULU, DISNEY+ | Entertainment | | SHELL, CHEVRON, BP, EXXON | Transportation | | AT&T, VERIZON, T-MOBILE, COMCAST | Utilities | | CVS, WALGREENS, PHARMACY | Healthcare | | VENMO, ZELLE, PAYPAL (person-to-person) | Transfers |

Important Notes

• Pattern matching is case-insensitive and matches against any part of the transaction description.
• Rules are applied in the order they were created. If a transaction matches multiple rules, the first match wins.
• Categorization does not overwrite transactions that already have a category unless you explicitly ask.
• Wilson stores rules in the categorization_rules table so they persist across sessions and apply to future imports automatically.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: openaccountant
• Source: openaccountant/skills
• License: MIT

Install and usage instructions live in the source repository linked above.