The assignment is to build a quick model to demonstrate to the CEO of an online gambling platform the potential of machine learning in detecting money laundering by customers. The CEO wants a basic demonstration model within a week.
The analysis is summarised in report.pdf, were we find 80% accuracy for the machine learning models. Given the small sample (1752 suspect players), we find tha deep learning (artifical neural networks) are not suited to this task.
Although all of the processing was done in the python scripts mentioned in the report, a simplified run-through is included as the Jupyter notebook PP+ML.ipynb