Fair Isaac – defining the top end of the data mining market

Written By:
Content Copyright © 2007 Bloor. All Rights Reserved.

It is obvious that data mining has become a far more mature and established technology in the last decade than its niche presence before. As the market matures so the players have established niches for themselves and offer differentiation from products such as SAS, that for many would appear to be the most obvious choice. Whilst the majority of players have moved down market from SAS, Fair Isaac have adopted a path of really defining the top end of this market, with a product offering which is capable of offering both models which, in the majority of cases, they would claim to be superior, as in more powerfully analytic than a SAS model, but which are also more productive, enabling better models to be produced and deployed faster. Whilst with SAS many of the deployment and productivity issues can be addressed by embedding their data mining product in their overall BI architectural offering, that obviously involves a cost which gives Fair Isaac an opportunity to compete very favourably.

Whilst Fair Isaac is clearly not going to be a low price option, that it is so well established with leading financial services players, and key bodies such as major Telcos, testifies that it is an affordable solution. Its precision and productivity more than justify the investment. Its niche market is going to be based on those highly regulated and highly visible (as in errors would attract the attention of not just the regulator but the media) sectors such as mortgage loans, credit cards and loans. But Fair Isaac has a broader appeal, being used by Telcos, Government agencies, retailers and the like, so it should not be looked at as only suitable for Finance. It is a product for those who require high precision and want to control their models with care to achieve reliable results.

The offering is based on three main components:

  • Firstly there is Model Builder, which starts the process by building the models. Model Builder supports the CRISP-DM lifecycle of analysis through to deployment. It is an integrated development environment that offers a wide range of different techniques. From my experience Fair Isaac offers a more sophisticated and subtle way to treat data set than any other analytics tool. It therefore will favour those with the skills to use those options.
  • Secondly there is Blaze Advisor™, which is the Business rules manager. The rules advisor enables the intelligence that is generated in the models to be harnessed in a workflow. With this software, rules can be deployed faster and with more accuracy ensuring that the business benefit is maximised.
  • Finally there is the Decision Optimizer, which enables the portfolio of models that can be used to define a modelled scenario to be evaluated to work within the constraints of the real world to produce the best results for a given set of circumstances. With most data mining tools optimisation works by finding the best fit model to achieve a business goal; with Decision Optimiser it is the best fit portfolio which is looked at. This is an important distinction that sums up a lot of the difference between the Fair Isaac approach and the majority of the competition. Whilst others look for a single best fit, Fair Isaac is looking at what combination of models offers the best solution. Fair Isaac may well use different models to handle different subsets of the data and in combination come up with a far more powerful solution.

Together these form an Enterprise Decision Management (EDM) environment that offers a very powerful and productive means to address the critical, high volume decisions which need to be made with precision and authority if business is to operate efficiently and effectively. As the enterprise requirement is to constantly make more, and better, decisions, faster such systems are an essential part of the armoury of sophisticated corporates. Fair Isaac can provide many examples of how their technology has been deployed to enable users to make significant competitive gains.

As we see the database giants moving to put data mining into the kernels of their databases and to all but give it away many may question why you would need something of this sophistication, but it should be remembered that the likes of Oracle, Microsoft and IBMs DB2 are general purpose databases and for analytical applications at the high end of the market they start to run out of steam. Fair Isaac is very well integrated with the Teradata technology that is widely used to store the masses of data that financial services companies, Telco’s and government bodies have to store.

I think that in a world in which commoditisation is growing at an accelerating pace, market leaders need to be able to store vast volumes of data and analyse it quickly, with great sophistication, and to then deploy the results with assurance to reap the benefits. For many areas of their core market Fair Isaac can offer preconfigured solutions, and for all areas of their market they have professional services of their own or from a network of market leading partners. The high quality and the innovation which lies behind Fair Isaac makes it a serious contender for such companies and is worth serious evaluation against any of the competitors.