IT Consultancy and software design ltd have since 1998 been involved in the highly specialized field of fraud detection and fraud detection systems. Increasingly, financial institutions and businesses do not include fraud detection as a part of their core business and look outside the company for the level of expertise required. There is a growing need for experts in this area; we are therefore offering our services internationally.

Ours main services can be categorized as follows:

Performance analysis.

All fraud detection systems will at some point in time need to be refreshed if they are used correctly. This is self evident, the system is build on historical data, and as new fraud patterns emerge, the system will increasingly be unable to cope. But when does this occur ?
We have over the years developed a method to measure the performance of fraud detection systems. The methodology used to detect fraud is not relevant here, it may use artificial neural networks, fuzzy-logic, rules or it may even be manual.
We help establish the guidelines and standards for the client that determines at what level the fraud detection system needs refreshing based on the number of frauds experienced.
The guidelines can then be compared to the values we calculate in the performance analysis, which include:
· False/Positive
· Card Detection Rate
· Transaction Detection Rate
· Volume Detection Rate
· Detection Delay

A typical performance graph:

Fraud analysis

Statistical research on your frauds is essential when determining how, if necessary to attack the problem. You need to know how much and how many, when, where and if there are obvious patterns. We deliver a detailed report and suggestions on steps to take.

Feature finding
For systems running artificial neural networks, we can find the features most likely to be affective. Each feature will be given in mathematical notation with a detailed description and its correlation coefficients with fraud specified. There may be anywhare from 50 to hundrads of feature in a given system A typical feature can be expressed mathematically e.g..

Rule finding
For systems running on rules, we can find the ones most likely to be affective. Each rule will be given in mathematical notation with a detailed description and its correlation coefficients with fraud specified. A typical rule can be expressed mathematically e.g.

Other Consultation
Fraud pattern research, fraud analyzes and rule or feature finding are fundamental in all fraud detection methods, we are also experienced in designing, developing and implementing fraud detection systems and/or assisting in choosing the right product for your organization.

Please do not hesitate to contact us if you think our services will benefit your company.

Email us at jonb@it-cons.com