Consumer banking behavior examples from the PKDD'99 Discovery Challenge dataset.
Financial transactions data are closely guarded by banks, which makes (anonymized) public datasets almost non-existent. This is unfortunate, since financial behavior has many interesting particularities. I've worked within banks in the past and often felt like sharing related insights.
Thankfully, a Czech bank did build such a dataset for a Data Mining challenge in 1999, and it seems that this remains - 20 years later - the only one. Also known as the "Berka dataset", I got my copy here (and put a backup here).
The Jupyter notebook transactions.ipynb
provided in the present repository contains examples of individual customer behaviors extracted from this dataset. A related transactions.html
export is available for easier viewing. To re-run the notebook or do your own analyses, please download the dataset either from where I found it or where I saved it afterwards.