THE SQL Server Blog Spot on the Web

Welcome to - The SQL Server blog spot on the web Sign in | |
in Search

Dejan Sarka

Fraud Detection with the SQL Server Suite Part 5

This is the fifth, the final part of the fraud detection whitepaper. You can find the first part, the second part, the third part, and the fourth part in my previous blog posts about this topic.

The Results

In my original fraud detection whitepaper I wrote for SolidQ, I was advised by my friends to include some concrete and simple numbers to calculate the return on investment (ROI) in a language that managers can understand. with some customers, we really managed to get very impressive numbers. However, I am not repeating this calculation here. However, this is my personal blog. Therefore, I am writing my personal opinion here.

I am kind of bored with this constant requests to show simple numbers that managers can understand. Personally, I don’t think that managers are that stupid that they would not understand anything beyond primary school mathematics. And even if some of them are that dumb, I don’t care, as they can’t become my customers. They would never be able to understand the value of such an advanced technique like data mining.

I am pretty sure that the vast majority of managers can calculate approximate ROI by themselves, and also better than I can do. They definitely know their business better than I can do, and already know how much money they are losing because of frauds and how many frauds they are already preventing or catching early. In addition, I am pretty sure that most of the managers do understand the value of learning, and appreciate building of the learning infrastructure.

Therefore, in short, I am leaving to you, to the reader, to evaluate what can you expect from implementing a fraud detection continuous learning cycle. And thank you for understanding my point!


Fraud detection is a very popular, albeit very complex, data mining task. I have developed my own approach to fraud detection. The most significant element of this approach is the continuous learning cycle.

Although Microsoft SQL Server is not the most popular tool for data mining, I am using it. The SQL Server suite gives us all of the tools we need, and because all of the tools come from a single suite, they work perfectly together, thus substantially lowering the time needed to bring a project from the initial meeting to a production-ready deployment.

Another advantage of my approach is the mentoring with the knowledge transfer. It is not my intention to get permanent consulting contracts; I want to progress together with my customers. Once we finish the project, or sometimes even as soon as we finish the POC project, the customer can begin using and continue improving the fraud detection system constantly with the help of the continuous learning infrastructure.

Finally, due to Microsoft’s licensing policies, the customers that already possess Microsoft SQL Server Standard Edition or higher and Microsoft Excel, do not need to purchase any additional licenses.

Comment Notification

If you would like to receive an email when updates are made to this post, please register here

Subscribe to this post's comments using RSS



Trevor Dwyer @sqlartist said:

Hi mate, just came across this series on fraud detection - I wrote an article/paper on insurance fraud using SQL Server, Data mining in Analysis Services - I used the descision tree algorithm and also to add a bit of depth to the piece I wrapped a third party neural network algorithm into SQL Server and compared the abilities of both. I was quite pleased at the time and I even had some professor cite it in his PHD thesis.

All the best

January 10, 2014 7:06 PM

Dejan Sarka said:

Trevor, nice to hear from you again. Hope you are doing well. Is this whitepaper published somewhere? A quick bingle search did not return results.

January 11, 2014 3:21 AM

Dusan said:

Hi Dejan,

Thanks for the great articles on the topic. My group is a SQL Server shop with very few non-Microsoft products in the pipeline. We deal primarily with fraud detection and have developed somewhat significant SSAS data mining experience. I have seen all the focus Microsoft has put into Azure ML and without having any features added to SSAS data mining, I wonder how much more development do you think somebody should invest on top of SSAS for data mining? What are your thoughts?

We are on-premisses and wouldn't benefit significantly from Azure ML so we are exploring integration with R through Revolution Analytics' DevelopR web services framework, for example.

Many thanks my Balkan neighbor,


January 3, 2015 1:53 AM

Dejan Sarka said:

Hi Dušan!

I would say that as long as SSAS Multidimensional is included in SQL Server suite, Data Mining will be there as well. MS is investing in Azure ML, right. However, this is not going to be free anymore. Therefore, it also depends on users, on us.

Thank you for the comment and best regards,


January 7, 2015 11:43 PM

Leave a Comment


About Dejan Sarka

Dejan Sarka, MCT and SQL Server MVP, is an independent consultant, trainer, and developer focusing on database & business intelligence applications. His specialties are advanced topics like data modeling, data mining, and data quality. He is the founder of the Slovenian SQL Server and .NET Users Group. Dejan Sarka is the main author or coauthor of fourteen books about databases and SQL Server. Dejan Sarka also developed and is developing many courses and seminars for SolidQ, Microsoft and Pluralsight. He is a regular speaker at many conferences worldwide for more than 15 years, including conferences like Microsoft TechEd, PASS Summit and others.

This Blog


Privacy Statement