Jump to content

Financial Crime Buster Analysis Template for Spotfire® 1.3

1 Screenshot


This Spotfire template guides the user through the tasks of adhoc data discovery, supervised model creation and unsupervised model creation to build a strategy for combating financial crime.


The approach used in this template for fighting financial crime places machine learning at the center of the crime detection system. Machine learning models use historic data to learn how to spot risky or abnormal behavior exhibited by transactions, clients, suppliers, or other players. It uses two types of models: Supervised learning algorithms, that tell us how similar to past fraud a new transaction is and Unsupervised learning algorithms, that tell us how odd a new transaction seems when compared to past transactions.  The first model guarantees accuracy, the second the ability to adapt to changing realities.


Existing financial crime solutions suffer from two problems:

  1. Many false positives
  2. Long investigation times

The curse of existing crime fighting solutions for financial institutions is that they are black boxes which require hefty consultancy fees to update.

At Spotfire, we can improve your situation in a number of ways:

  • Apply machine learning to your historic data in a fully transparent approach
  • Leverage your big data without moving data across and without coding!
  • Make machine learning uncomplicated so your business people can interact with it directly
  • Compare new models with existing ones to make sure you always improve
  • Do what-if analysis to adjust your investigative efforts to the size of your team and vice-versa
  • Understand how good a model is, which KPIs contributed most to it, track model versioning
  • Deploy models ultra fast with Spotfire Streaming
  • Keep track of all alerts and all investigative actions taken on them, such that everything is auditable at any time in the future
  • Justify why any transactions reiceived a certain treatment
  • Verify that all investigators are following expected procedures, identify bottlenecks, manage workload

Why it matters and how it works

 (4.5 minutes)

Customer Successes

Release v1.3

Published: May 2017

  • TERR scripts in the SendToStreamBase_CC_Supervised and SendToStreamBase_CC_UnSupervised data functions now facilitate debugging of the real-time scoring function.  See the embedded scripts for details.  
  • Now handles previously unseen categorical values when scoring the random forest model. Those values are now labelled "Unseen".
  • UI Improvements


Release v1.2

Published: November 2016

  • Improved Random Forest function.
  • "Send to Streambase" function now allows sending code to be run in real time.


Release v 1.1

Published: October 2016

PCA improvements


Release v1.0

Published: October 2016

Initial Release

  • Create New...