The objective of this project is the development of a Machine Learning service for the detection of suspicious activity patterns in user accounts. In particular, the software has the following characteristics:
- Automatic segmentation of user transactions for the identification of specific behaviors.
- Identification of user behavior changes that indicate an account take over.
- Built for big data, scaling effectively to real-time streams of data.
- Runs on client’s premises, providing an easy to use predictive models API.