/CreditCardStatementParse

Trying and failing to parse and classify credit card transactions

Primary LanguageJupyter Notebook

# Parse Credit Card Statement into Data


# Setup

## Parse
If you happen to have a CapitalOne credit card you can download your monthly statements as PDFs and run capital_one.py.  
See `./capital_one.py -h` for options.

Example:  
`./capital_one.py ./assets/CapitalOne/Statement_112023.pdf -p 3,4`


## Analysis

The goal here is to classify transactions by the short text descriptions included in each transaction description.    
Very much a WIP since my attempt at zero-shot classification doesn't work very well.  
I'm working on hand labeling some of the data for some supervised learning.  


Some scripts in the analysis dir use the OpenAI API. 
You could use them by adding your openai key as environmental variable "OPENAI".  
If using `virtualenv`, edit <your-env>/bin/activate and add `OPENAI=your-key` at the end of the script.  
Also add `unset OPENAI` inside of the `deactivate ()` function in the activate script.