Originaly forked from https://github.com/robintw/BankClassify and the modified to match my needs

conda create --prefix ./envs python=3.9

conda activate /Users/jangie/Projects/DataScience/BankClassify/envs

docker run -p 8888:8888 -v $(pwd):/home/jovyan/work -v /Users/jangie/Documents/Finanzen/Detailed_Data:/home/jovyan/data jupyter/scipy-notebook

raw exported data is put into the raw folder. The the function add_raw_data reads in new data, checks for duplicates with existing data and appends new data to the clean data, where each year is kept in a individual file for each source. The classifier is then trained on the clean data.

data
├── raw
│   └── paypal
        └── raw_data_1.csv
        └── raw_data_2.csv
        └── ...
│   └── another_source
        └── raw_data_1.csv
        └── raw_data_2.csv
        └── ...
├── clean
│   ├── 2020_paypal.csv
│   └── ..