Get a hands-on introduction to advanced methods of detecting fraud, security threats, suspicious behaviour and other anomalies, using data science, predictive analytics and other statistical, computational and visualisation techniques. These methods provide ways to detect “unknown unknowns” where simple rule-based methods and computer-assisted audit techniques (CAATs) fail.
The course provides key concepts and with hands on practice with a range of readily available and free tools, including Microsoft Excel and R, a powerful open source data analysis tool. It is suitable for all practitioners in fraud detection, law enforcement, security, compliance, insurance, audit and the finance function seeking an introduction to and hands-on experience with data analysis techniques.
Who should attend?
This course is suitable for all practitioners in fraud detection, law enforcement, security, compliance, insurance, audit and the finance function seeking an introduction and hands-on experience with data analysis techniques.
It is also perfect for IT and data analytics practitioners seeking to add fraud detection capability to their existing analytics skill set.
This course will provide learning at a number of levels. At the conceptual level, the course will cover key fraud and anomaly detection tools, and teach their main strengths, weaknesses and other distinguishing features.
Participants will be exposed to a range of rule-based, statistical and visual tools for detecting fraud and other anomalies.
Hands on exercises will provide participants with experience in the actual application of methods presented to data including real-world examples.
A range of tools will be presented in detail along with exercises to provide participants actual experience in detecting anomalies in data.
Simple methods such as rule-based Computer Assisted Audit Techniques (CAATS) will be presented first. These are tools for detecting well-defined, known anomalies. These are simple to understand, simple to implement methods widely used in accounting and audit practice. These simple methods serve as a natural staring point, and their weaknesses a motivation for more advanced methods.
The remainder of the course will be on advanced statistical and visual techniques, including Digital Analysis, Predictive Fraud Detection and Multivariate Outlier Detection, Association Rules and Social Network Analysis.
These are advanced analytics techniques, and many of these are also used in Data Mining / Predictive Modelling / Big Data applications outside of fraud / anomaly detection. These techniques are presented as powerful ways of detecting “unknown unknowns”, anomalies that cannot be characterized in advance, and are detectable by their statistical signatures rather than business rules. Where CAATS deploy defined rules, the advanced analytics methods can actually identify new rules and enrich future CAATS libraries.
Participants will also be exposed to powerful data visualisation techniques to support their analysis of suspicious outliers, falsified figures, as well as Social Network Analysis for detection of collusion and time series analysis.
Example exercises will be conducted in Microsoft Excel, and R, a powerful, free open source data analytics tool.
Accompanying theory will also describe real-life scenarios where these tools would be applied, and how the information provided fits into the broader fraud detection, prevention and investigation processes. This addresses cooperation between analytics and other members of a fraud or anomaly detection team, including subject matter experts, operational staff and investigations. Issues covered will include the necessary protocols and levels of understanding between these parties.
Theory will also cover the effective combination of CAATS with advanced tools, and the ongoing enrichment of CAATS rule bases with new rules discovered by advanced methods. This is an important knowledge management capability, providing growth in organisational “wisdom”. The rule base can learn from and deal with previously seen anomalies.
The course will be instructed by Presciient managing director Dr Eugene Dubossarsky. Eugene is a founder and Fellow of the Institute of Analytics Professionals of Australia (IAPA); Director, University of New South Wales School of Mathematics and Statistics Industry Advisory Board; and a recognised industry leader in Business Analytics. Eugene is an experienced, professional data miner of 20 years’ experience programming in R and its parent language, S.
Attendees are recommended to have completed Presciient’s Introduction to R two-day course, or equivalent.