« May 2006 | Main | July 2006 »
June 26, 2006
Eligibility management - a perfect opportunity for business rules
One of the most common uses of decisioning technology in Government is the management of eligibility. As you will see this is mostly a rules and compliance issue, though there may be a role for analytics in deriving some of the rules. In an eligibility situation you are typically trying to make it easy to do several things:
- Manage the rules of who is eligible for what
- Make it easy to change these rules as regulations change and court rulings are made
- Empower those who might or might not be eligible to find out without having to call someone – to "self serve"
- Regulate transactions based on eligibility, either by allowing/disallowing them or by executing them in the context of the regulations.
These characteristics make it clear how business rules are an ideal technology. For example, a large retirement association has created an online site that will allow government employees to check their benefit status, explore other benefit opportunities, and submit requests. The application is for pension fund management and handles more than 100,000 members. All the domain knowledge for pension administration has been externalized from legacy code to business rules, allowing those who understand the rules tomanage them. These rules determine benefits eligibility, calculate benefits, and validate data allowing for members to self-serve with the transactions for which they are eligible.
Similarly a US State wanted to allow parents to determine online whether or not their children are eligible for state-sponsored healthcare benefits. The business rules power a health care benefits portal - users log on to the site and enter in specific information. Rules then analyze the user's entry to determine benefit eligibility and costs. This health care benefits portal is available online 24/7/365 making it easier for parents to find out what their eligibility is and to carry out relevant requests forpayment. The rules themselves are easily maintained by non-technical business people who understand the complex layers of regulation involved, ensuring that the portal stays up to date and accurate.
Although eligibility is a rules-based problem, descriptive analytics can sometimes be useful in deriving business rules for use. For instance, a regulation might require that a certain percentage of those applying be allowed a benefit and descriptive analytics could be used to determine a set of rules that would identity the top (or bottom) x percent. Descriptive analytics might also be used to consider the impact of the rules as more and more court rulings are insisting that rules can be biased if their effectis biased even if the rules themselves have no specific bias coded into them.
Share:
Posted by James Taylor at 12:28 PM | Comments (0)
June 23, 2006
Podcast on business rules
http://edmblog.fairisaac.com/decisons_podcast/ has a new podcast on business rules. Enjoy.
Share:
Posted by James Taylor at 4:33 PM | Comments (0) | TrackBack
Healthcare and credit-based decisioing
Saw this interesting little piece in Health Data Management today - Hospitals Should Know the Score. An interesting snippet on how important proper credit management is for hosptials - "As much as 30% of patient payments end up being written off as bad debt". How can an EDM ,approach help solve this? Well one of the most important measures providers can take to improve their financial condition is toensure they are billing appropriately for the healthcare services they provide—an effort that starts at the front door. During the admissions process, rules-driven applications can perform real-time validation of patient-supplied data against medical records and external data sources. By making certain patients are who they say they are and that addresses and other information are complete and correct, providers reduce their financial risk.
Prediction models and external data sources enable an EDM solution to predict, at admission or pre-registration, what the optimal initial payment request for patients based on their particular financial situation, what the optimal settlement amount is, and how to fashion a payment plan with the highest likelihood of completion. Guidance regarding the initial payment request increases revenue upfront and improves the effectiveness of intake staff. Knowing what patients can and will eventually pay enables financialcounselors to help patients who need help and minimize the number who go to collections, while also generating maximum revenue for the provider system. This is similar to a usual business approach except that you don't want to decline care because someone can't pay so much as help them pay for the care they need. Rules can also simplify and speed up admitting by guiding intake specialists through the most efficient process for each incoming patient.
Another part of the solution is to employ rules and models at the point of care ,to perform identity checks and likelihood to pay. Whenever a patient will incur an out-of-pocket responsibility above a pre-defined threshold (e.g., $25), an EDM solution can inform providers of an optimal strategy to maximize the likelihood of payment. The solution can also check the patient's information against all rules governing eligibility for charity care, Medicaid, and all types of supplemental funding. Rules would becreated and deployed consistently throughout the provider system to comply with both government and organizational patient payment policies, and can be monitored for compliance by staff throughout the organization. This prevents unecessary bad debt being accumulated during a patient's stay.
Once a patient is discharged, an EDM approach can also be used with overdue accounts to improve collection results and minimize recovery costs. Instead of treating all overdue accounts with the same sequence of dunning letters and calls, providers may, in fact, be able to collect more money by doing less. Analytics can identify differences between accounts that affect payment behavior—segmenting first-cycle delinquent accounts, for example, into those likely to self-correct, those likely to be influenced by collectionstreatments and those unlikely to pay under any circumstances. Providers can use this segmentation to save money by making fewer outbound contacts and thereby also reducing the volume of inbound inquiries such contacts generate. They can make an early referral to a collections agency of those accounts with very high nonpayment scores.
An analysis of delinquent accounts and collections work streams using an experimental design can enable a provider organization to determine which tactics are most effective for which types of accounts. An EDM solution can route accounts in collections to the work stream that will produce the most revenue in the shortest amount of time at the lowest cost.
By applying EDM to the problem of bad debt throughout the patient lifecycle, hosptials can collect more of what they are owed more quickly without having to turn away patients. Fair Isaac is developing an offering, Patient Decision Manager, to do just this and partnering with some leading hospitals to make it work.
Thanks to my colleague Larry Feinstein for help on this one.
Share:
Posted by James Taylor at 11:35 AM | Comments (0)
June 22, 2006
Using business rules for Anti-Money Laundering
Interesting report by Henry Peyret of Forrester on A Business Rules Engine Decreases The Cost Of Fraud Detection. The system he describes is a classic use of a non-inferencing rules engine but I want to make a couple of points of clarification:
- Most business rules leaders, including Fair Isaac, allow for both inferencing and sequential execution. Depending on the rule scenario one or the other runs faster. There is some discussion of when you need inferencing here and here (as well as Wikipedia)and a description of what the newest versions of Rete (Rete III) can do here.
- Sequential execution is always going to be faster in terms of rules evaluated. It spends no time figuring out which rules to fire and so fires more. This works great unless you need to find a small number of rules to execute amongst a very large number of potential rules in which case inferencing will be better. This is why good rules products allow for both.
- We have clocked Blaze Advisor, using a real case study from a Fortune 100 company, at 3,543,424 rules per second per CPU or over 6.5M predicate evaluations per CPU per second. Per second. 10M exchanges per hour would be easy. If we assume 30 rule execution per transaction (implied in the paper) then this is 300M rules per houror 5M rules per minute or less than 100,000 rules per second.
And remember that Blaze Advisor does its code generation under the covers so you could change the rules in the production system using a web-based rule maintenance application and not have to do any kind of re-generation/re-compile step. This avoids the problem Henry notes of making it hard to re-deploy changes. - As Henry also notes not all rules lend themselves to trees (or tables for that matter). That's why good products like Blaze Advisor and JRules allow for multiple representations.
AML and other kinds of simple fraud detection are great for rules. Even if you have more complex fraud, requiring analytics, rules are still a crucial component in a solution.
Share:
Posted by James Taylor at 9:33 AM | Comments (0)
June 21, 2006
Integrating Analytics into Business Processes
In The Four Styles of Integrating Analytics Into Business Processes ,Gareth Herschel and ,Betsy Burton of Gartner discuss the ,four kinds of analytic integration with business process – intra- or inter-application and explicit or implicit. Comparing this to my model of enterprise decision management it is clearthat I am talking about analytic integration with an implicit calling mechanism. For the kind of analytic integration I am discussing the inter-/intra- distinction is less important and can be considered a technical decision. Here's how they describe implicit invocation:
"Analysis that is implicitly called by the business application may be invisible to end users, constraining their choices or displaying relevant data as they follow the business activity."
This is how analytics are used in EDM applications – the analytics drive rules or decisions or are part of how a decision is made. They are typically not requested by a user. As far as the user is concerned they are requesting a decision, it is the design of the application that causes the analytic to be embedded. This is not to say that explicit embedding of analytic is not useful – far from it – it is just typically notwhat I mean when I talk about analytics in EDM.
So let's consider the two types that seem most relevant when thinking about EDM – these are what Gartner calls “Style 1: Intra-application and implicit” and “Style 3: Inter-application and Implicit” In both cases the analytic services are automatically performed without explicit user prompting but in the first case they are part of the primary application, in the second part of a different application (probably delivered through a service-oriented approach).
The key impacts of the various types are discussed in the paper so I will just make a couple of points that seem to be key for EDM:
- “Business users see the results of the analysis or rule-driven recommendations as a seamless part of their business applications.”
This is the core premise of EDM – the role of analytics in EDM is to improve the quality of decision, not to make business users (or customers self-serving etc) decide they need to know something. The attraction of this approach is clear – the need for the user to have analytic skills or the time to apply them is elilminated. They can, in Gartner's words, “focus on the rapid use of the analysis rather than its creation”. In particular this offers users of the application analyticinsight regardless of the user's skill in data analysis and this can be key when considering self-service applications and those aimed at low-level/high-turnover staff such as those in call centers. - Analysts working on developing these kinds of analysis have an additional responsibility. They can no-longer say “I built a great model now it's someone else's problem”. Instead it is their job to make sure that they understand the context in which the analysis will be used and that the implementation of their model is reasonable and cost-effective. A good model that is actually running in a real system beats a great model no-one canimplement.This is one of the ways in which analytics in an EDM sense is different from some traditional data mining/analytic approaches. It really matters how easy and quick and accurate the process of turning the analysis into "code" is.
- For the IT folks, the model development process often involves huge amounts of data with all that implies for performance. However this should impose only modest constraints on IT as the number of analysts is small and the process of building the models is batch-oriented even if done often. The operational impact of executing the analytics in the system should be fairly small as only rules and executable “equations” will need to berun with high performance and throughput. Again this is a key feature of EDM analytics - the complexity is in the development of the analytic, not its execution. A predictive scorecard is a classic example - lots of data must be processed, complex algorithms applied and sophisticated techniques used but the end result is wonderfully simple to execute.
Companies that have implemented a business rules platform to make decision automation easier will have no trouble adding analytic insight to those decisions using this implicit invocation approach. Companies trying to embed analytic insight into traditional code have only themselves to blame…
Share:
Posted by James Taylor at 4:10 PM | Comments (0)
Don't be adversely selected against - use EDM!
I saw this article in National Underwriter - Data Use, An Edge For Some Insurers: Consultant. In this the authors of the study say
"Property-casualty companies that reject traditional business models in favor of the creative use of data mining and client segmentation will have an advantage in the coming years"
This reminded me of the discussions I have with insurance customers around being adversely selected against and how decision management, especially decision management that takes advantage of data mining and predictive analytics, can help. So what is "being adversely selected against" I hear you say. Well consider two companies, one has three prices (for good risks, moderate risks and bad risks respectively) and the other has nine (very good, somewhat good, slightly good, very moderate, somewhat moderate etc). This means that if a specific customer is in the good tier for company 1 they might be in any one of three tiers for company 2. Now company 2 is smart - they make their price for their "very good" segment better than company 1's "good" segment, their "somewhat good" segment price about the same and their "slightly good" price a little worse. Let's consider three potential customers:
- Customer A is a very good risk and so gets a better price from company 2 (who puts them in their "very good" segment) than from company 1 (who puts them in "good")
- Customer B is a somewhat good risk and so gets about the same price from the two companies
- Customer C is a slightly good risk and so gets a worse price from company 2 than from company 1
Given this, customer A is more likely to pick company 2, customer B might go either way and customer C will likely pick company 1. Repeat over many customers and company 1 is likely to end up with more customers like C and fewer customers like A. Instead, therefore, of having customers spread evenly across the segment company 1 will have more customers at the low end of the range ("somewhat good") than at the high end ("very good"). This in turn means that the average customer for company 1 will be somewhere between "somewhat good" and "slightly good". In contrast, company 2 will get more "very good" customers and an average between "somewhat good" and "very good". For company 1 this means they are being adversely selected against - they have priced their segment based on the average but are getting people who average worse than that. If your competitors have more fine-grained pricing than you do, this is what happens. You end up with a customer portfolio that tends to be worse, on average, than you expect.
How do you avoid this? Well by more finely targeting and segmenting your customers. This takes analytics or data mining. If you can use your data to better predict risk and then combined this risk assessment with the rest of your data before segmenting your customers then you can get more fine grained segments. The two graphics show how more segments help you better match the reality of a continously changing risk curve. There's more on segmentation here and on the insurance industry and EDM here.
The key, as the study's author said, is to take informed risks.
Share:
Posted by James Taylor at 9:55 AM | Comments (0)
Don't be adversely selected against - use EDM!
I saw this article in National Underwriter - Data Use, An Edge For Some Insurers: Consultant. In this the authors of the study say
"Property-casualty companies that reject traditional business models in favor of the creative use of data mining and client segmentation will have an advantage in the coming years"
This reminded me of the discussions I have with insurance customers around being adversely selected against and how decision management, especially decision management that takes advantage of data mining and predictive analytics, can help. So what is "being adversely selected against" I hear you say. Well consider two companies, one has three prices (for good risks, moderate risks and bad risks respectively) and the other has nine (very good, somewhat good, slightly good, very moderate, somewhat moderate etc). This means that if a specific customer is in the good tier for company 1 they might be in any one of three tiers for company 2. Now company 2 is smart - they make their price for their "very good" segment better than company 1's "good" segment, their "somewhat good" segment price about the same and their "slightly good" price a little worse. Let's consider three potential customers:
- Customer A is a very good risk and so gets a better price from company 2 (who puts them in their "very good" segment) than from company 1 (who puts them in "good")
- Customer B is a somewhat good risk and so gets about the same price from the two companies
- Customer C is a slightly good risk and so gets a worse price from company 2 than from company 1
Given this, customer A is more likely to pick company 2, customer B might go either way and customer C will likely pick company 1. Repeat over many customers and company 1 is likely to end up with more customers like C and fewer customers like A. Instead, therefore, of having customers spread evenly across the segment company 1 will have more customers at the low end of the range ("somewhat good") than at the high end ("very good"). This in turn means that the average customer for company 1 will be somewhere between "somewhat good" and "slightly good". In contrast, company 2 will get more "very good" customers and an average between "somewhat good" and "very good". For company 1 this means they are being adversely selected against - they have priced their segment based on the average but are getting people who average worse than that. If your competitors have more fine-grained pricing than you do, this is what happens. You end up with a customer portfolio that tends to be worse, on average, than you expect.
How do you avoid this? Well by more finely targeting and segmenting your customers. This takes analytics or data mining. If you can use your data to better predict risk and then combined this risk assessment with the rest of your data before segmenting your customers then you can get more fine grained segments. The two graphics show how more segments help you better match the reality of a continously changing risk curve. There's more on segmentation here and on the insurance industry and EDM here.
The key, as the study's author said, is to take informed risks.
Share:
Posted by James Taylor at 9:55 AM | Comments (0)
June 19, 2006
Top posts on BPM, SOA, CEP, BAM and business rules
Continuing the analysis and got this list is of ones related to business process, SOA, CEP and related topics. Here goes:
- Are BPMS and BRMS complementary or not?
There are a number of posts on this topic including this one based on my Brainstorm appearance and this one (on the other blog) on not over-synchronizing processes and rules. - Windows Workflow Foundation - rule engines and rule management
Biztalk, Windows Workflow Foundation and Business Rules
These two posts discussed Microsoft's technologies, their ability to do rules and how a rules management system would fit with and enhance both WWF and Biztalk. - The future of BPO includes business rules
Business Rules and resisting the commoditization of process
Some articles on how and why business rules might make sense when considering BPO and standard processes. I wrote more about this here ,and here. , - Podcasting on SOA and decisioning
Is Orchestration tailor-made for business rules?
These two relate to a popular topic - the use of business rules in SOA. There's a whole section on this on the blog as well as some posts on the other blog - SOA, BPM, CEP and business rules
Complex Event Processing
The use of business rules in event processing and related topics comes up often. There is a section on the blog ,on event processing with additional posts. ,
Share:
Posted by James Taylor at 10:25 AM | Comments (0)
Top posts on business rules
I was doing some analysis the other day and identified your favorite posts and I thought it would be fun to show the list with some comments. The first list is of ones related to business rules. Here goes:
- Agile Rules? A "conversation" with Scott Ambler
This post summarizes some discussion I and others had with Scott (a leading thinker on agile methods) on the role of business rules in agile methods - What is Rete III?
A basic overview of the Rete III or optimized Rete algorithm invented by Charles Forgy as an improvement over the original Rete algorithm. - What are the main benefits of business rules technology?
Why are business rules better than traditional code?
Basic overview of the benefits of business rules management systems and some specific suggestions as to why business rules can be better than code. The blog section on business rules has more on this kind of topic. - How do business analysts maintain business rules?
A discussion on one of the hottest topics in business rules - how, exactly, do business users manage business rules in production applications. The blog has a lot on business agility and I wrote on how rules enable collaboration here also. - How do I build a list of Business Rule Engine Selection Requirements?
Some notes on how to identify the requirements for a business rules engine or business rules management system - Are there any standards under development for business rules
Question on business rules standards. There is more here ,where I identify the two main standards in play. - Writing Better Requirements - Key to Success or false hope?
One of my favorite topics - why trying to "fix" requirements is not going to solve the business agility problem. There is lots more on this topic here. - Business Role changing?
A posting by Barb von Halle that provoked a fairly spirited discussion on how business rules might be impacting the role of business. - Starting with Legacy Rules
Barb again on legacy rules and getting started with them. There's more on the blog on legacy modernization.
Share:
Posted by James Taylor at 9:18 AM | Comments (0)
June 17, 2006
Top posts on going "beyond BI"
Last list of top posts. This list is of ones related to business intelligence. Here goes:
- Predictive Analytics and Insurance
A post on how predictive analytics can be used for better decisioning in Insurance. There's more on this in the Insurance section ,and on the other blog here. - Live from Gartner BI: Future Trends in Business Intelligence
Live from Gartner BI: Business Intelligence, Business Process and Business Rules
The differences between BI and EDM
Some sessions on trends in BI and how they point towards EDM as an approach. The use of a business rules platform to enable the embedding of analytics into processes is a key topic here - operational BI as it is sometimes called. There's more on this on the other blog. - The Harvard Business Review focuses on Decisions
This was an interesting issue of HBR and highly recommended. - EDM: A platform for strategic decisions
Rahul Asthana wrote this nice piece pointing out that just because EDM focuses on operational, high-volume decisions does not mean it cannot and does not deliver on strategic alignment. It does.
,
Share:
Posted by James Taylor at 10:14 AM | Comments (0)
June 15, 2006
Insurers should collaborate to fight fraud - with data
Saw this article To Beat Fraud, Insurers Must Collaborate: NICB and it hit home. Collaboration is part of how the credit card industry drove down fraud. By pooling data on usage and fraud they were able to enable to building of sophisticated fraud models. These neural nets, combined with business rules in Fair Isaac's Falcon Fraud Manager product, delivered an EDM solution that had dramatic results. 
I have blogged about the need for insurers to do the same before.
Share:
Posted by James Taylor at 9:29 AM | Comments (0)
June 13, 2006
Customer Analysis when you have millions of customers
This interesting article on Customer Flashcards caught my eye yesterday. Chris and Zach do a nice job of presenting the idea of customer flashcards, combining lots of information into a single image, and how that might help you see more complex patterns and recreate a feel of customer intimacy (about which I have blogged before). I really liked theconcept and their examples were fascinating. What made me pause, though, was thinking about how to apply this kind of approach when I have hundreds, thousands or even hundreds of thousands of customers. Am I really going to scan that many flashcards? The discussion of Blink was also interesting but again made me wonder about applying this kind of "thin-slicing" (at which humans excel) to huge datasets (something at which humans do not excel).
So how could you apply this kind of approach to huge numbers of customers? It seems to me there are two ways:
- Firstly you could apply the approach to customer segments. After segmenting your customers using a data mining approach, you could then display the various aspects of each segment using the flashcards. This might help turn the statistical view of the data mining tool/analyst into a more business-focused flashcard, enabling the business folks to better understand the segments that were statistically significant.
- Secondly you could take the approach of using flashcards as part of identifying the kinds of variables that might make sense in a predictive scorecard. For instance,%26nbsp,a cardholder risk scorecard will typically use information like credit history trends and patterns as%26nbsp,part of the data analyzed to produce a risk score. This essentially automates for a system the multi-variable/patterns/trends combination presented so nicely in a flashcard.
Like all visualization techniques this has great potential for helping people make decisions. Equally, like all visualization techniques, it becomes challenging when your volume gets high enough to need decision automation so that, for instance, your website can do as good a job deciding as the people who can see the flashcards..
Share:
Posted by James Taylor at 1:07 PM | Comments (0) | TrackBack
June 12, 2006
Book Review: The Real-Time Enterprise
This book was recommended to me by Mike Gonzalez over at Hands-on-BI. The book is a couple of years old but its time horizon was fairly long so it is still nicely relevant. The book introduces the concept of a Real Time Enterprise, discusses how exactly you can compete on time and talks about how various approaches, especially business process management, can and must be used as part of moving to an RTE. The book tries to avoid too much hype but sometimes falls victim to overblown rhetoric nevertheless.
There is a lot in the book about operational transformation, clearly a key ingredient in moving to an RTE, and the authors do a good job covering the people and culture issues as well as the technology ones. Some specific thoughts:
- The book puts some emphasis on codified rules or business rules but tends to lump them as part of describing a business process, something with which I disagree.
- It uses many examples of real-time business decision-making like cross-sell and fraud detection that are classic EDM targets
- It makes the point that you need organizational knowledge to be "explicit, executable, actionable and adaptable" - a clarion call for a business rules management system if ever I heard one.
- The authors focus on "right time" - the responsiveness needed to add value or create competitive advantage.
- There is a nice little example that perfectly illustrates why decisioning is so important in this kind of situation. They discuss how an RTE responds to a crash of a truck carrying monitors by finding other comparable monitors and sending them. To me this is full of decisions as well as process - what is a comparable monitor for this customer? How urgent is this delivery? how willing are customer and supplier to incur additional cost? and so on.
- I think the concept of Decision Yield would fit well into some of their discussion of how to measure processes.
- There is an interesting chapter on an event-driven company and how Complex Event Processing or Business Activity Monitoring is key in an RTE.
- Where the authors talk about "cycle time" I might have spent some time talking about the decision cycle and the competitive advantage to be gained by changing decisions faster and more effectively than your competitors.
- Finally they talk about having change built in to systems - agility being something I have blogged on before.
There is a great quote from Dr Michael Porter of HBS who , said "The essence of strategy is choosing to perform activities differently than rivals do". For me, the essence of strategy can be choosing to decide how to treat customers differently than rivals do. That takes EDM.
It's a readable book and a nice overview of some of the trends and responses in real time enterprises. There is more on the blog on several of these topics and the book can be purchased here.
Share:
Posted by James Taylor at 4:22 PM | Comments (0)
Different approaches to adding insight to your decisions
I got an interesting comment ,the other day asking how to effectively conduct data mining in an EDM context - outsource or not? This made me think it would be worth posting a general description of the different ways to add analytic insight to a decisioning environment. Assuming we are talking about adding predictive analytic models to operational system (see this article for a discussion onpredictive analytics), there are several ways to bring analytic insight to bear on a decision.
- Find a score
Clearly if there is an externally available score with a good reputation and relevance then use that. This is by far the cheapest and most effective - you don't need data or data mining/analytic expertise, just an understanding of your business and an ability to match it with a score. If, for instance, you are trying to manage credit risk then the FICO score or its global counterpart Global FICO is a great tool. These are typically available from bureaus (in the US) and sometimes have different names (BEACONand EMPIRICA are both FICO scores, for instance). - Find a score that seems to be a good substitute
If, as in most decisioning problems, you cannot find a score that works exactly consider if any of the ,scores that are available might be a good surrogate. Larry Rosenberger (Fair Isaac's head of R&,D) gave a great keynote on this at InterACT recently where he talked about the fact that many scores can predict unexpected things - for instance, your FICO score turns out to be a reasonably good predictor of whether you will stay fit or not.No-one is suggesting that there is a causal link here - having a good FICO Score does not MAKE you fit - but that some underlying characteristics (responsibility or ,integrity perhaps) mean you tend to act in ways both that drive up your FICO score and that increase your likelihood of fitness. You can thus analyze your customers by comparing behavior to available scores to see if any are a good (non-causal) predictor. A word of warning - don't ever rely on a single score, especially a singlescore for which you don't have causal linkage, to make a decision on its own. This will not work well and will get you in trouble with regulators (and quite rightly too). - Work with a third party DSP
There is growing interest in something called a Decision Service Provider or DSP. This involves basically outsourcing the decision-making piece of your process. You make the issue of finding the right data, working out where and how to get it and then decisioning with it someone else's problem. ,The DSP hosts the common and regulatory rules (helping you be compliant) ,and allows you to customize for your businesswhile handling the complexities of assembling a rich data set and analytics that use that data set. By and large this works best when there is lots of third party data that might make sense for you to include, e.g. about consumers, but where you don't have relationships with all the data providers. - Have someone analyze your data and send you models (not reports)
This is the classic way to get started. Work with someone like Fair Isaac who has the analytic/data mining skills to turn data into insight. You make your data available, though hopefully they have expertise too that will be brought to bear, and they help you by cleaning and arranging the data to make it usable to build predictions and then by building a predictive model from it. You want someone to hand over an actual model, or an analytically derived set of business rules, rather than a report that just tellsyou what you should do as it is important to keep the time delay between getting data and using that data to make better decisions to a minimum. Ideally you will use a package or build into your EDM framework a place to put the model so as to enhance your operational decisioing. You will likely also want to have the model refreshed on a timely basis (what is timely varies depending on the model). - Have someone analyze the data AND teach you in the process
You can take the approach in 4 and adjust it by having it done on-site with tools you can buy yourself. If you combine this with training and having your staff act as second-seat behind a qualified modeler you can build your own team and skills in parallel.This works great if you have a desire to staff up and do this yourself long term but want to get short term results. - Buy tools and training and do it yourself
There are good tools out there for doing the analysis and building models from it and even a few, like Fair Isaac's Model Builder, that will deploy the model for you into production systems. You can hire staff, train them and develop a unique brand of analytics for your decisions based on your customers.
Share:
Posted by James Taylor at 2:16 PM | Comments (0)
Shifting performance management into action
I wrote this article for ebizQ on "Shifting Your CPM Into Action". EDM offers a platform for decisioning across processes that enables you to change the way your company works as your performance management infrastructure tells you that you should. Check out the various categories on this blog for more on these topics and don't forget my blog over on ebizQ - http://www.ebizq.net/blogs/decision_management.
Share:
Posted by James Taylor at 1:29 PM | Comments (0)
Self-regulation and decision automation
I got an interesting comment the other day from Craig Cameron about the potential for self regulation - the fact that some businesses want not only to comply with government enforced regulations but also to "comply" with the wishes of socially conscious consumers, for instance. This got me thinking about how this might be the same or different from other kinds of ,compliance-oriented decision automation. Some thoughtsthen:
- Like most compliance problems automation is primarily a rules problem.
Showing that every transaction had followed the same rules, being able to show which rules fired in which transactions, managing exceptions effectively and so on would all be key, even if the rules were things not to do with regulation but with "social" compliance. - There is a growing role for analytics
Just as we see in regulation a trend towards expecting companies to manage the statistically likely outcomes of their rules, I would expect to see the same in social compliance. If the effect of my rules is to have a socially unacceptable impact overall, it will not matter that the rules themselves seem OK. - Rules may need to be more visible in this case
Unlike a typical compliance problem where the rules will likely only be shown to regulators or auditors, rules for social compliance might have to be published to show the public they were being followed. This would make the normal readability benefit of a business rules approach even more critical. Being able to display the actual rules and have an auditor, say, confirm that they are what is running in the system might be critical. Having a readable syntax would be key. - Customization of rules by consumers or trading partners
If a company starts offering customers (whether consumers or trading partners) options for social compliance, they will likely need to allow a degree of customization. A company or consumer might, for instance, want to specific a maximum additional cost for a renewable component let's say. This will mean using the power of a business rules management system to mange layers of rules - central ones, local ones and customer-centric ones.
Thanks to Craig for an interesting line of thought.
Share:
Posted by James Taylor at 12:07 PM | Comments (0)
June 9, 2006
Identity Theft - some stats and suggestions
I got a comment from the folks at Metrics 2.0 asking about stats on Identify Theft (something about which I have blogged a couple of times - here and here). I asked around and got some information. According to Javelin Strategy &, Research, Identity Theft affected about 9.3million people in 2004, causing $52.6 billion in losses in the U.S. alone. This was reported on in BusinessWeek. Since then Javelin has updated the report for 2005 and losses climbed to $56.6B (a fairly slow rate of climb). This compares with some $3 billion in losses for credit and debit cards combined for the US. The huge difference between these numbers has led some to criticize the Identity Fraud numbers but remember ,itdoes include things like mortgage fraud, utilities fraud, telecom fraud, auto loans, HELOC, etc. One identity theft can result in lots of losses across these different channels. That said I suspect the real identity fraud number is lower than $50B but still much higher than the credit/debit card losses.
So if you are worried about Identity Theft, and you probably should be, what can you do?
For companies
- Think about automating identity theft protection - the combination of business rules and predictive analytics can be very effective at eliminating or reducing this kind of fraud
To see how effective this approach can be, take a look at how well it worked in reducing credit card fraud in the US - down by over 65% since the introduction of Falcon Fraud Manager, which uses rules and analytics to detect and manage credit card fraud.
- Remember that many feel that identity theft losses are overstated - don't get distracted from other problems just because Identity Theft is "hot". Think about fraud more holistically - this is the basis for Fair Isaac's approach with Falcon One.
- Think about all aspects of how you manage fraud identification decisions - see this post for additional thoughts.
- Remember also that fraudsters will go where you are not watching - push down in one place (with better procedures, more technology) and they will pop up somewhere else. This is a war not a battle.
For individuals
- Don't Panic! The number of cases is declining and more than 2/3 of victims do not incur any out of pocket expenses.
- Most victims know how their data was stolen and it happens offline, not online so worry about losing your personal details everywhere not just online.
- There are some more good tips on the BBB site or you can go to Identity Theft Resource Center, the government's site ,or sign up for a service like the myfico Identity Theft product
Share:
Posted by James Taylor at 3:56 PM | Comments (0)
June 8, 2006
Compliance, business process and business rules
I wrote a short article on the use of Business Process Management Systems (BPMS) and Business Rules Management Systems in compliance over at BPM Institute recently.
There's more on this blog on compliance and Business Process Management as well as some specific posts on why to keep them separate ,and on how to bring analytics into processes using rules.
Share:
Posted by James Taylor at 9:17 AM | Comments (0)
June 6, 2006
CEP needs business rules too
I saw a comment on IT Analysis that talked about The difference between complex event processing and event stream processing. I liked the distinction drawn but have to take issue with the comment Philip makes agreeing with a CEP vendor about "why and where CEP works and BPM and rules engines don't. "
Its customer is an airport and its application is baggage handling...Clearly you could model this using a BPM process. However, what you would be modeling is a perfect day. ...—so the key to baggage handling is dealing with exceptions. Further, the exceptions are always different and have to be managed in real-time, which is why you need a CEP engine
Now I agree that this is why CEP complements BPM - the BPM engine handles the normal state and the CEP engine can handle exceptions. Where I disagree is in the idea that you don't need a business rules engine or a business rules management system. In fact I would go so far as to say that the fact that there is a case to be made for both BPMS and CEP is precisely why you do need a business rules management system!
Managing the events in real-time is clearly a specialty CEP capability. Deciding what to do about these exceptions is a business decision. The normal, day-to-day business process will also need some or all of those business decisions. Partner airlines or other external systems as well as other internal processes might want to use those decisions and decisions about routing baggage might depend on non-baggage related rules like those agreed in contracts with partner airlines on treating gold customers.
Keeping these business decisions in a business rules management system would allow them to be managed and re-used effectively across these different systems. Embedding them in the CEP system just means repeating them one more time. On top of which you add yet another place where you might want to start applying predictive analytics instead of just being able to analytically enhance the business rules environment and so benefit everyone.
Just because a product can handle business rules locally does not mean it should.
There's more on why BPMS and BRMS should be separate here and in Separating rules and process on Sandy Kemsley's ebizQ blog and much of the logic applies to CEP too.
Share: