Out of the Recession

-- Posted by Carole-Ann

It has been a little over a year since Lehman Brothers went down.  Although the recession is technically over, we all know that we have a long road in front of us for the economy to get back, for unemployment to go down and for consumers to feel confident again.

Many industry experts have had a chance to comment since then on what happened and what we should do now to get out of it.  I stumbled upon some interesting reading that I definitely recommend.

What triggered our interest at FICO was the sudden abundance of papers from industry analyst firm Gartner on a new concept called Pattern-Based Strategy.  It looked like a new trend was emerging across many independent disciplines.

Pattern-Based Strategy

Pattern-Based Strategy resonated with me from the get-go because it aligns beautifully with the Decision Management concepts we have been advocating for a few years.  I am not saying that we had already invented the wheel of course -- I am not that arrogant!  I was rather impressed that our Decision Management perspective fitted quite well within this framework, if I can use that term.  On the same token, very different perspectives such as Human Resources (which is the industry my husband develops applications for) happen to fit beautifully as well.  Outside of our nest, I tried to challenge the concept with all the industries or disciplines I knew quite well, and I inevitably came up with the realization this was applicable and it was likely a winning strategy.  This is to some extent a business model of the survival of the fittest strategy.  Given my background in biology, how could I not be drawn into it?

Survival of the fittest

Survival of the fittest relies on two main concepts: constant changes and natural selection based on best adaptation to the environment.  In pattern-based strategy, the mechanism removes some of the randomness of the changes.  Gartner advocates to proactively keep your senses alert, on the lookout for new trends, so that you can adjust your behavior and measure performance to make sure you will fit better in the new environment.  From that standpoint, it feels like pattern-based strategy is like an "aware" survival of the fittest strategy.  It is clear that genes do not have any "awareness" -- they change by chance or by design "to change", there is no intent of a better state.  Selection is the only mechanism that will elect who makes it and who doesn't, assuming that the change is discriminant.  The active pattern seeking is obviously an addition to the survival of the fittest theory -- a brilliant one.

In the new few days, I will go over the various aspects of pattern-based strategy and link those to my favorite subject: Decision Management.

DM Tools User Group is a Success!

-- Posted by Carole-Ann

What a week it has been!  We got fantastic presenters like Jim Joy from Avnet and Uma Kandasamy from ACS presenting Medicaid Claims Processing, an amazing roster of marquis companies like Oracle, American Airlines, Honeywell, Chubb Insurance and Discover Financial Services, as well as the most acclaimed industry experts like Steve Hendrick (IDC), James Owen (InfoWorld author and chair for October Rules Fest) and Neil Raden (Hired Brains).

Arriving at home, I needed a break though, a little escape from thinking.  While I was wrestling with my sewing machine, it occurred to me how much frustration or satisfaction can result from the performance of the supporting staff.  The more I sew, the more I appreciated the worry-free organization of the User Group.  Don't take me wrong I never underestimated their efforts and skills...

I could have used LinkedIn to post a recommendation for each one of them but since they are such an important part of the success of FICO, I thought it was worth making it in the EDM Blog.

First the project management at large.  We switched responsibilities mid way through and eventually benefited from both being engaged.  Michael Small and Jenn Rigley have been the number one reason we got everything done on time, what I called a worry-free process...  Worry-free from my perspective of course!  I am afraid of being personally responsible for quite a few white hair on their heads!  Without losing their smile they accommodated every single one of my last minute eccentric requests such as finding an ESPN Monday Night Football mp3...  Yes, there is a perfectly good reason for needing that for one of my presentations.

For as long as I have been with Fair Isaac / FICO, I have been totally impressed with our Events team.  Every single detail is anticipated, addressed and perfected.  No surprises (except for a birthday cake at our last San Paulo roadshow).  The show looks great thanks to this team.  There is no other word than kudos to this very professional team: Chrissy Ruiz and Kristen Hinsberger.  The rest of the team is actually as good too, especially my good friend Joy Thomas.

I could go on and on with those acknowledgment but I would surely miss someone since this has been a fantastic team effort.  Content leaders, expo experts and Monopoly table captains have been superb.

Thinking back about my weekend miseries, in the end, I was pretty happy with my new curtains but I will try to forget how much swearing went into it -- believe it or not but none of that came out while organizing the UG ;-)

Show Season for BRMS and Decision Management

-- Posted by Carole-Ann

 We are all getting ready for a lot more traveling this end of year.  Although I don't like being away from the kids that much, I do like the opportunity to meet customers and other BRMS or Analytics practitioners.

I I have a compiled a little list of shows where we may meet this year.  Feel free to contact me if you'd like to meet in person!

First of all, our User Group!  We have an all time high attendance level for the FICO Decision Management Tools User Group.  We will meet in Chicago in just a few days - September 16-18.  If you have not yet signed up or if you want to check the latest updates in the agenda, feel free to check out website: http://www.online-reg.com/FICO/2009/dmtools/.  I highly recommend that you come and listen to Steve Hendrick from IDC.  His keynote will be quite interesting!  Marquis companies will talk about best practices in Business Rules, Predictive Analytics and Optimization.

Early October, I will also attend the Gartner BPM Summit in Orlando -- http://www.gartner.com/it/page.jsp?id=911413.  Down there, I would also recommend that you track down Jim Sinur and David McCoy if you are into Decision Management or more focused on BRMS.

On a more general note, the Forrester Business Technology Forum 2009 will be held in Chicago -- http://www.forrester.com/events/eventdetail?eventID=2383.  Key analysts to track there will be John Rymer and Mike Gualtieri.

One of the biggest shows for anything IT will be held in Florida at Gartner IT*Expo -- http://www.gartner.com/technology/symposium/2009/sym19/home.jsp.  The focus will be on balancing costs, risk and growth.

At the same time, October 20-21, we will sponsor Predictive Analytics World in Washington DC -- http://predictiveanalyticsworld.com/.  I will not be able to join but Andy Flint will represent FICO Product Management and would be glad to meet with you.

October 26-30, Hard-Core Rules Specialists will meet at October Rules Fest in Dallas, TX -- http://www.octoberrulesfest.org/.

For more business-focused BRMS presentations, you can join us at Business Rules Forum from November 1st through the 5th-- http://www.businessrulesforum.com/.

Last show in November will be the FICO Blaze Advisor Insurance User Group, at our customers' initiative in the Northeast (Connecticut).  As soon as the date is finalized and details published, I will share them with you all.

CEP a Predictive Modeling facility? Myth!

-- Posted by Carole-Ann

Despite my efforts to simplify the Decision Management picture, especially as it relates to CEP, BRMS and BPM, it seems that the CEP buzzword turned into an omnipotent ruler of the Decision Management world, at least in some discussions...

I have been exchanging a few tweets these past couple of days on this very subject with John Rymer, Mike Gualtieri and a few others.  It became eventually clear that 140 characters were not enough...

Is CEP the best solution for predicting future behavior?

That's pretty much what got me started on this Twitter thread...  My first reaction was amazement and incredulity.  Why would anyone think that????

I am not going to go into the Predictive Modeling basics here.  In order to create a predictive model, you need data.  This seems to be generally admitted.  In order to assemble a good data sample you need to put in place some mechanism to collect this data from your applications.  CEP is one of the techniques being used to collect.  This is not the only one -- keep in mind that Bill Fair and Earl Isaac came with the brilliant idea of creating risk models in 1956 when they founded FICO and they could do it without CEP ;-)

The goal of modeling is to deploy your predictive models of course.  Today Enterprises use a BRMS such as Blaze Advisor to import their models (potentially from PMML -- Predictive Model Markup Language) and use them to score applicants or customers against risk, propensity to purchase or any other dimension you might be interested in.  Please note here that they might do it themselves or have their solution provider do it for them transparently.  Others implement the models directly into their databases or into the low-level code.  One other possibility is to use CEP to deploy those predictive models.

Let me develop a bit those aspects.

CEP and Event Correlation

Complex Event Processing is a technology that receives data input (events) and has an opportunity to do something about it, hopefully really fast.  You can squeeze there some logic to correlate events.  The exciting part is that the system keeps some memory of the past so you can reason over time "if I received more than X alerts in the past hour...", "when this follows that...", etc.

CEP is one technique, there are other implementations of similar behavior.  One could use BPM to do the very same thing.  I would be concerned about performance though.  When you go down the architecture layers, at the broker or bus layer, performance is much higher.

At FICO, we used our own C or C++ implementations for demanding Telecom systems like Roamex.  Call detail records (CDR) come into the system real-time and need to be correlated.  The mediation system modifies the events to collate 2 CDRs that start and end at the very same time for the same origin and destination for example.  We are talking about huge volumes and extremely high performance needs.  We could possibly have used CEP but at the time it was not performant enough.

Anyway, the point is that we should not confuse a usage of technology with one of its implementations.

CEP and Data Collection

I talked about data collection being a critical component of preparing for predictive modeling.  Since CEP knows how to make sense of data coming in, CEP can be used to stream the application data into a datawarehouse.  It would be an overkill but I can see this taking off if the collection logic becomes more complicated over time.  With always increasing volumes, future systems may not want to collect everything and post-process, reduce and clean up the application data.

A similar usage is Business Activity Monitoring (BAM).  You need to be able to extract data fast to expose valuable insights to business owners.  A BAM system could take advantage of CEP to subscribe to data but it does not make CEP a BAM system.

Again, the point is that you can use CEP or you can use of mechanisms.

CEP and Predictive Models

Predictive modeling is obviously our forte at FICO.  We deploy those via various methods: as part of Decision Management Suite (within top-of-the-line Apps such as Falcon for fraud detection or TRIAD^TM for account management), within credit bureaus (any way they want it) or custom when models are developed with Model Builder.  We've seen our custom models going into all sorts of environments over time.

Deploying a model within a CEP system would make sense.  Mike cited examples of stock trading monitoring apps.  The reason they can be well-suited is that the time-sensitive logic applied at the time the event is received could very well be issuing red flags according to a predefined predictive model.  Keep in mind that predictive models are typically a mathematical expression in the end -- not a complex ruleset.

For fraud detection, adaptive models are totally appropriate as the fraudsters keep coming up with new ideas that you want to prevent.  Falcon uses neural nets that learn from consortium data as well as adaptive routines that learn yet more rapidly.  CEP could theoretically been used but because of the more complex logic to be applied and high performance needs, we have opted for our own BRMS and profiling capabilities.

The point again is that CEP is one of the possible techniques but not the only one.  In some instances, it is actually more performant to use another implementation.

Does CEP technology itself predict?

As far as I know it does not.  Our Blaze Advisor BRMS can apply predictive models but it is not a predictive technology per se.  Model Builder is.  I would apply the same logic here.

Could you build an adaptive system using CEP and BAM?  Of course.  CEP could feed BAM and BAM could potentially update real time the logic within the CEP system.  Because CEP is used inside doesn't mean that CEP is the predictive technology.

CEP = Decision Management?

Let's play with logic here...

To prove that A=>B is false, I will prove that (not B) does not => (not A)
A = Decision Management
B = CEP

Sorry I could not resist...  Let's put it in simple terms.  Can a Decision Management application not involve CEP?

Going with real life examples:

• Falcon: Fraud Detection
• TRIAD: Account Management
• Capstone: Originations...

That was easy of course...  But let's focus on one use case in account management a minute.  Those big Financial Services institutions run batch processes nightly to assess who should receive a credit line increase, who should be sent to a collection agency, etc.  There is no triggering event.  The bank just gets ready for the next day so that their call centers, collection teams or marketing department knows what to do with each one of their customers.

No triggering event, no event to process, just data management.  Decision Management also covers those aspects that are totally foreign to CEP.

Definition, definition, definition...

We often run into similar confusion at FICO.  Because one application or component embeds one technology, it does not mean that the underlying technology is equal to the higher level system.  Xpress is not Decision Optimizer because it is embedded in there.  Blaze Advisor is not TRIAD.

CEP has value and CEP can be used for Decision Management. 
CEP is not Decision Management just because it is a part of it.
Let's not be victims of a buzzword fashion!

I am expecting heated reactions.  Let's discuss...

Predictive Analytics are hot!

-- Posted by Carole-Ann

We just hosted an introductory webinar on Predictive Analytics and I was amazed at the turn out.  BRMS practitioners are now getting interested in mass...  They are engaging with FICO into education, discussions, etc.

Why?

I never doubted that it was a good idea to mix and match predictive models with business rules.  It adds an element of Precision that you don't get out of traditional business rules.  Regulations and Expert judgment can't be mined of course but there is a good deal of intelligence that resides in those past transactions.  You can find out what your customers are likely to buy, which ones are likely to go delinquent or to fraud.  If you are better at finding those people than your competitors, you get an unfair advantage.  You can offer red-carpet treatment to those customers that will be greatly profitable in the long run, and let your competitors fight for those that will cause collection fees and other types of losses.  If both companies use predictive models, the one with the best insight will stay ahead of the other one, assuming of course that the strategy that uses that insight is sound!

So I wasn't so much interested in the why than I was in the why now....

All the signs in the economy are indicating that we are getting out of the recession.  We are going to suffer for a little while because the unemployment rate will linger a little longer, especially with so many people not being able tor retire.  The US are in debt.  They are not the only country in the world...  It is going to be a long way back up but many analysts seem to agree that we have reached the bottom and should expect growth from now on.

We have learned quite a bit in the process.  We have learned about:

• the importance and value of predictive models -- Not having predictive models suitable for the new types of financial products increase the financial institutions' exposure to unprecedented risk levels (for more, read http://www.edmblog.com/weblog/2008/11/the-role-of-predictive-analytics-in-the-sub-prime-crisis.html)

• the lack of transparency in the processes can lead to diluted accountability -- This is no great surprise to anyone that bailout money comes with increased duties in terms of transparency and audit.  Accountability is a joke when your name is not associated with any specific enterprise strategy that can be measured; it is not any more serious when, as a manager, you do not have visibility and control over the process or the results.  Getting a better insight on the expected results of a policy change (minor or major) and having to sign up on this change in the way the whole company is doing business will lead to better enterprise performance

With those lessons learned, it suddenly makes sense that Enterprises pay more attention now than ever to Predictive Analytics in the context of Business Rules.

I am glad to see the change happening in the marketplace.  Financial Services have been ahead of the curve in terms of using those Predictive Analytics, especially in the context of business policies.  It is interesting to see increased interest in many different industries.

Decision Management Tools User Group

-- Posted by Carole-Ann

We decided to host a User Group this year for DM Tools products: Blaze Advisor, Model Builder and Xpress for Optimization.

Given the economy and lack of travel budget we expected a reasonable turn around for the show, maybe a bit below our past performance.  What a mistake!!!

Maybe customers realize how impactful Decision Management Tools can be for the enterprise.  Maybe they are looking for even better best practices for optimizing their current usage.  Whatever the reason might be, we have seen record-high registrations this year. 

In response to the high demand we have had to upgrade.  We signed up for larger facilities, we put away our SOLD OUT sign and now accept more registrations.  This is going to be awesome!  I am looking forward to meeting all of you at the user group.  It is a wonderful opportunity to share and learn.  I am excited to see our vision coming through with more and more DM practitioners combining one or more tools in their apps.

See you in Chicago this Fall!

October Rules Fest 2009

-- Posted by Carole-Ann

My abstract has been accepted at October Rules Fest!  I will be presenting as well as my fellow Chief Architect Carlos Serrano Morales.

I will be talking about the feedback loop and business performance monitoring.  Carlos will be focusing on the popular subject of Cloud Computing.

As a reminder, the show is in Dallas the last week of October 2009.

If you have not yet registered for the event, feel free to do so on their website at http://www.octoberrulesfest.org/

There is also a LinkedIn event you can sign up for and let everyone know you will be there: http://events.linkedin.com/October-Rules-Fest/pub/79911

Business Rules or Spaghetti?

-- Posted by Carole-Ann

In the early days, we use to think about Business Rules as a great way to extract logic from the code.  The vision of several nested if-then-else statements evoked the type of spaghetti code we all wanted to get away from, and we did.

Now that Business Rules Management Systems (BRMS) have made it to the typical application blueprint, the fear and pain of maintaining spaghetti code is well behind us...  But should it?

Don't take me wrong, I would never even dream of saying "let's go back to those days".  I have a passion for Decisioning Technologies and really appreciate the benefit it brings to the table.

But what if we created a new type of problem?  Now that complex, really complex business rules can be created with ease, is it possible that we are toying with new limits?

I have seen some very large systems with tens of thousands of business rules.  Granted you don't have to worry about the inter-dependencies.  But you still have to figure out what rule you need to update, what context it refers to.  With tens or hundreds of rules, not a problem.  With thousands it become more challenging.  If you have even more business rules, it may become a hard problem.

A customer I worked with a little while ago showed me how they navigated a world-wide repository of business rules.  I was amazed by the time they spent clicking around to find where some piece of logic were located (country- or region-level for example), and make sure that they applied the right rating table for the right population of end customers.  Navigating, selecting, browsing, validating, updating...  What could have been a simple update, in the end, relied on many error-prone user-maintained naming conventions and a lot of time wasted navigating what could have resembled Business Rules Spaghetti!  No blame on the customer, they did what they were supposed to but the nature of business rules, regardless of the technology you use, opens the door for those issues.

If you allow me analogy here, it is like your file system.  It is not bad at all (I don't know how we could do without one actually) but when you manage a 500Mb or 1Tb disk with tens of thousands of files, you are bound to click around to find the picture you archived a while ago.  This is why Google desktop and other search techniques are so useful.

Back to the Business Rules world, this is what happens with maturity.  You end up with large volumes and eventually complex dependencies that may be contextual (when editing rules related to product ABC you want to pick one of the Pricing tables for Product ABC, other pricing tables are irrelevant).

We have a formidable opportunity now with new approaches and technologies to better handle content.  Instead of thinking "the old storage way" about structuring a rules repository, why not adapt like Google, and think in terms of tags, dynamic structures, dynamic context and dynamic filtering.  This is elegant and flexible.

I am impressed by the potential of such technologies applied to BRMS.  The customer I talked about is able to boost the productivity of the business analyst or business users in charge of the maintenance with very little changes to their rules.  Actually only the structure of the repository and a few data providers changed.  Dynamic filtering and linking turn an old Rules Maintenance Application with growing pains into a slick and empowering tool.

Live from InterAct - preshow tutorials

-- Posted by Carole-Ann

We flew into NY on Monday.  After what seemed a long day of internal meetings and presentation rehearsals, we started the show in the afternoon with our tutorial.  I was part of the "Operationalizing Analytics" track.

Scott Horwitz talked about the how Insurance companies can benefit from Business Rules, Decision Simulation, Predictive Analytics and Decision Optimization.  If you have been reading this blog, these should not be new concepts to say the least.  It was good though to engage the crowd on their current practices.

I was responsible for the Decision Simulator part and demo.  It was refreshing to see heads nodding when I talked about existing customers' experiences.

Horia finally showed a demo on Decision Optimization.  Customers had lots of questions on regulatory compliance.  I anticipate that we will get more and more compliance questions over time, especially as the Obama Administration defines new rules for the Insurance business.

Before the Welcome reception, managment briefed the whole staff about the name change.  We are now FICO, rather than Fair Isaac.  The official marketing count of attendees for the show is over 500.  Not so bad given the economy.

Now, let's get ready for tomorrow's big day.  I have 2 presentations.  I must say that I am pretty excited about my NFL presentation.  A French woman discussing football in New York.  It has to be a first!  If you happen to be in town, don't miss it.

Now introducing Decision Simulation

-- Posted by Carole-Ann

Et voila!

We are now delivering Decision Simulation to the BRMS world.  What an exciting day!

Let me give you some insight as to what that means in reality.  Up to now, the industry has been focusing on what we call "validation" capabilities.  Validation is about running your business rules against test cases, sometimes historical data, and assess how it compares to expected values.  You end up with a listing of pass versus fail transactions.  Typically you just care about the discrepancies (what failed).

Decision Simulation is a key component of the Decision Improvement loop I always talk about because it offers a real alternative to Business Users.  Why do they need an alternative?  Because validation tells them IF the new business rules (I often call it a strategy) are what you expected them to be, it does not tell you HOW they will perform, in business terms of course.

In the world we live in now, we cannot afford to make sub-standard decisions out there.  If you fail to price new applicants properly, you will lose business.  If you grant too much credit to risky applicants, they will not be able to repay and you will lose money.  If you process fraudulent claims, you will lose money. If you lose money, you may not be able to stay in business.  We see examples every day that lead to bailout plans, mergers and acquisitions, or worse.

Being able to assess the value of your decisions before they hit production is valuable.  It may be the most valuable thing to the business owner: boosting his confidence that he is making the right decision.  How do you define right versus wrong?  It is really up to you: it could be measured in terms of revenue, profitability, risk, etc.  With a good set of Key Performance Indicators (KPIs), business users can chart the business impact of their strategy changes.  It assumes of course that your recent historical data is a good enough predictor of what is yet to come.

The tool that we are releasing now provides the kind of insight to the business users that Fair Isaac has provided to its leading customers, mostly in Account Management via TRIAD or our custom Strategy Science engagements that leverage Decision Optimizer.  Decades of expertise embodied into a set of out-of-the-box reports.  It is really exciting to finally make it available to all of our customers!

If you are an existing Blaze Advisor customer, I will be glad to host you on our Introductory webinar tomorrow (Wed 12/17 noon EST / 9pm PST).  If you are not, we will have a session in January that does not assume knowledge of our leading BRMS.