Syndicated from ebizQ

John Reynolds had an interesting post a little while back where he shared some thoughts on Perfect development tools. His emphasis was on support for things like iterative and test-driven development but it seems to me that there is also a need to move application development beyond code.
While developers do need development environments that support new approaches to developing code that works and that help speed and improve the application development process, they also need tools that help them move beyond code. In particular they need a development environment that:

• Ensures that decision-making logic is managed declaratively as a set of business rules
• Integrates analytics into this decision-making logic in a useful way
• Helps them put process or workflow into a process management tool
• Helps them define events and how events will be correlated and processed

As long as development environments assume that everything can and should be written as code, I do not believe they will be “perfect”. Code is the right way to do a whole bunch of things in application development but organizations are discovering that new tools for managing events, processes and rules are effective at increasing agility and bringing business users into the application evolution process. Similarly the importance of analytics is only growing. Excluding application developers from these trends is in no-one’s best interests. A perfect development environment should support these concepts and integrate it with tools for effective code development.

I have blogged before about Application Development 2.0 as well as the power of business rules in model-driven development/agile and the ways in which decision management affects the Software Development Lifecycle.

Syndicated from BeyeNetwork

In a great post on 8 things to keep in mind on predictive analytics, some folks from Diamond Management & Technology laid out some things to keep in mind that I really liked. Here they are with my comments – you can get more detail on each from the series of posts with which they followed this initial one.

1. Understanding the cost of a wrong decision helps target investments
   Absolutely, though I still think that finding a decision you can tie to an executive’s compensation plan works better.
2. Strategic and operational decisions need different predictive
   modeling tools and analysis approaches
   .. and deployment approaches. I divide decisions into strategic or direction-setting ones, tactical or day-to-day management ones and operational or transactional ones. Particularly with the latter, which are crucial, you need to think about how the models will be deployed if they are to add value.
3. Integration of multiple data sources, especially third-party data,
   provides better predictions
   Yup, but don’t just integrate your data – begin with the decision in mind and integrate to support it.
4. Since statistical techniques and tools are mature, by themselves
   they are not likely to provide significant competitive advantage
   True. It is their ability to turn YOUR data into YOUR insight that does.
5. Good data visualization leads to smarter decisions
   .. at the strategic and tactical level and to better models at the operational level – decision making at the operational level is too high-speed, too automated for much in the way of visualization to be useful a the moment of decision.
6. Delivering the prediction at the point of decision is critical
   Yes!
7. Prototype, Pilot, Scale
   Of course – don’t forget to scale the deployment piece too
8. Create a predictive modeling process & architecture
   Yes. And map it to your IT development process if you want to impact operational decisions embedded in your enterprise IT infrastructure.

A great list!

Syndicated from ebizQ

I am working with some folks at SAP on a new BPM book – Applying Real-World BPM in an SAP Environment. I am working on chapters about the role of decisions in processes (check out this post for some help on this topic) and on the use and management of business rules in automating decisions. For those of you working with SAP NetWeaver or BRFplus I think and hope you will find the book really useful.

If you want more information, download the RealWorldBPMPostcard.pdf

Doug Henschen has a great interview with Deepak Advani of IBM, the new head of IBM’s newly acquired SPSS business (and I am not just saying that because he mentions Smart (Enough) Systems).  I am looking forward to seeing what IBM does with the combination of ILOG and SPSS, along with InfoSphere, WebSphere, FileNet and Cognos.

Eugene from SARS, the South African Revenue Service, presented next on how SARS is using BI in revenue collection. He began by pointing out that there is a difference in how public sector organizations use BI – a focus on service delivery not profits, on taxpayers not customers, enforcement campaigns not marketing campaigns and so on. Of course public sector organizations still want an ROI, operational efficiency and use KPIs for performance management.

SARS has a wide range of core systems as well as a set of external data sources. Initially the IT department just dumped data from the source systems to their business users. This was replaced with a more formal information management department that responded to requirements defined by analysis teams but still hit the source systems. Capacity constraints led to an enterprise data warehouse (Teradata) but the Information Management department could not meet the demand for new reports etc while the business users wanted more control. Their current state is that of having their information management department acting as an enabler for business departments to manage their own BI capabilities. The technical architecture behind this has a primary staging layer for moving data into a production warehouse Operational Data Store and a secondary staging area supporting BI and data mining warehouses. This two stage approach allows them to present historical data through the lens of constantly changing business rules. A metadata repository underpins this and a presentation layer gives users access to reports, cubes etc.

SARS presents strategic summaries, aligned with the KPIs, as dashboards for the executive level who are typically considered measurement users. Tactical reports and dashboards are delivered to regional offices. These users tend to be exploratory users. Finally operational intelligence is delivered to execution users at the operational, branch level. The different levels consume different kinds of analytics.

SARS has learnt not to pursue big bang projects, to mix business and IT people, to plan for poor data quality and for peak season volumes and to manage change. From a business perspective they focus on changing how business people request data/reports, on showing ROI and on embracing user empowerment and self-service.

They use standard reporting on things like ontime filing, with an ability to drill down into zones, industries and more as well as self-service for reporting on metrics against various dimensions, slice and dicing etc. More interestingly they use various advanced analytics to catch fraud etc. For instance, a company might under report its corporate income tax and over-report the VAT it paid so that it continually gets refunds. However, this is a challenge because:

• Some critical fields are not mandatory
• It can be hard to correlate these two kinds of tax return
• Suspicious activity may have been reported but it is purely unstructured text.
• At the end of the day the intent is to find those organizations who are truly suspicious so data on registration, status, payment rates/timeliness must also be considered.
• And not everyone can be pursued so who to call and who to audit.
• Finally, are there linked entities that need to be closed down when a fraudster is found.

Advanced analytics are used in various ways:

• Neural nets predict values, or at least buckets of values, for missing values
• Statistically infer outliers
• Text mine the unstructured text reports to see if there are patterns of reporting that will allow early investigation
• All of this feeds into a risk engine that predicts the risk of fraud
• They then predict who is likely to be reached by the call center to prioritize calls to these taxpayers
• Next they predict the likelihood of a successful audit so that the auditors can prioritize their work
• They use association and geospatial data to find clusters of suspicious organizations, linking directors, audit companies etc.
• 3rd party information is brought in on things like houses and assets, travel etc to find suspicious mismatches between tax returns and lifestyle.

Great example of advanced analytics to detect fraud and catch tax evaders.

Eugene from SARS, the South African Revenue Service, presented next on how SARS is using BI in revenue collection. He began by pointing out that there is a difference in how public sector organizations use BI – a focus on service delivery not profits, on taxpayers not customers, enforcement campaigns not marketing campaigns and so on. Of course public sector organizations still want an ROI, operational efficiency and use KPIs for performance management.

SARS has a wide range of core systems as well as a set of external data sources. Initially the IT department just dumped data from the source systems to their business users. This was replaced with a more formal information management department that responded to requirements defined by analysis teams but still hit the source systems. Capacity constraints led to an enterprise data warehouse (Teradata) but the Information Management department could not meet the demand for new reports etc while the business users wanted more control. Their current state is that of having their information management department acting as an enabler for business departments to manage their own BI capabilities. The technical architecture behind this has a primary staging layer for moving data into a production warehouse Operational Data Store and a secondary staging area supporting BI and data mining warehouses. This two stage approach allows them to present historical data through the lens of constantly changing business rules. A metadata repository underpins this and a presentation layer gives users access to reports, cubes etc.

SARS presents strategic summaries, aligned with the KPIs, as dashboards for the executive level who are typically considered measurement users. Tactical reports and dashboards are delivered to regional offices. These users tend to be exploratory users. Finally operational intelligence is delivered to execution users at the operational, branch level. The different levels consume different kinds of analytics.

SARS has learnt not to pursue big bang projects, to mix business and IT people, to plan for poor data quality and for peak season volumes and to manage change. From a business perspective they focus on changing how business people request data/reports, on showing ROI and on embracing user empowerment and self-service.

They use standard reporting on things like ontime filing, with an ability to drill down into zones, industries and more as well as self-service for reporting on metrics against various dimensions, slice and dicing etc. More interestingly they use various advanced analytics to catch fraud etc. For instance, a company might under report its corporate income tax and over-report the VAT it paid so that it continually gets refunds. However, this is a challenge because:

• Some critical fields are not mandatory
• It can be hard to correlate these two kinds of tax return
• Suspicious activity may have been reported but it is purely unstructured text.
• At the end of the day the intent is to find those organizations who are truly suspicious so data on registration, status, payment rates/timeliness must also be considered.
• And not everyone can be pursued so who to call and who to audit.
• Finally, are there linked entities that need to be closed down when a fraudster is found.

Advanced analytics are used in various ways:

• Neural nets predict values, or at least buckets of values, for missing values
• Statistically infer outliers
• Text mine the unstructured text reports to see if there are patterns of reporting that will allow early investigation
• All of this feeds into a risk engine that predicts the risk of fraud
• They then predict who is likely to be reached by the call center to prioritize calls to these taxpayers
• Next they predict the likelihood of a successful audit so that the auditors can prioritize their work
• They use association and geospatial data to find clusters of suspicious organizations, linking directors, audit companies etc.
• 3rd party information is brought in on things like houses and assets, travel etc to find suspicious mismatches between tax returns and lifestyle.

Great example of advanced analytics to detect fraud and catch tax evaders.

Alan Fish has another great post over on his blog – DRA: Rules Discovery. He makes what I consider to be an essential point that is easily forgotten – you are not trying to find rules just so you know what they are, you are finding rules so you can make better decisions! Check out this post on the differences between decisions, rulesets and rules for a little more on this. This is part of what I call decision discovery.

I have just opened ITWeb’s BI 2010 in Johannesburg talking about decisions and importance of decision making in making BI matter (I will post my slides later). Great audience, nearly 200 people with a strong showing from end user customers (75%) and, very interestingly, nearly half considered themselves business / IT straddlers which is a great trend.

The next session, although focused on BI, was very relevant to decisioning and decision management. Martin Rennhackkamp spoke on strategic BI and focused on a number of critical things that help you make BI more strategic:

• Understand the strategic business objectives, make sure you understand them and focus on how BI can support them
  This is absolutely critical in my opinion. If you don’t understand your business and how your BI investments can support it then you cannot possibly make good investments in BI. I particularly liked how he mapped business objectives to BI initiatives, showing how much of what the objective needs is delivered by each. This allows you to see which ones matter and where you have holes.
• Focus on crucial business processes
  Again, good advice and I liked his focus on operational processes. Not only should people identify the data that flows through those processes but they should also ask if the processes getting the information they need to make the right choices, the right decisions. Especially, are the people executing these operational processes getting what they need.
• BI to the masses
  It is critical for information to flow down to everyone on the floor – pervasive and operational. And important not to buy the tools without thinking about how this works.
• Master measure management
  Liked this phrase. Bring KPIs into the picture, linking KPIs to operational behavior and data. Understand how different roles need to consume information and analytics and how all this maps to company objectives and measures. Focus most of your effort on measures that the consumer of a measure has control over and to which they contribute. As I said once perforce, a dashboard should do more than just raise your blood pressure!

eBureau is a predictive scoring and information service provider founded in 2004, focused on technology for very rapid model development and deployment. Using their own purpose-built modeling software, a small group of modelers developed 900 predictive models in 2009 alone. The company has been applying this capability for real-time and interactive marketing like contact centers, consumer lead generation, risk and fraud management, and display ad targeting. Not only can the models be built fast, they can be deployed in the cloud quickly for real-time scoring applications. Typical transactions take less than a second round trip. They have found that this approach works in a variety of industries like education, financial services, automotive, telecom etc. For instance, online universities are using the eBureau solution to predict which consumer leads will apply, enroll, and stay enrolled.

At its core, eBureau is focused on new customer acquisition whether helping clients understand payment risk or propensity to respond to an offer. They take historical performance data (leads, who converted, how valuable they were) and data from 50 other sources before running their predictive modeling technology. eBureau develops a predictive score (for fraud, probability to convert, payment risk, etc.) that can then be used to:

• Improve the online marketer’s cost-per-lead advertising decisions e.g. buy or no-buy decision on leads or right-pricing based on the score
• Improve contact center conversion efforts e.g. offer path management or how to route calls most effectively based on a consumer’s profile
• Improve display ad decisions e.g. find an audience that looks like your best customers and target the right creative at the right time using predictive models

eBureau has some 50 databases enabling them to cross-reference data like addresses and phones, customer purchase data like aggregated catalog sales data, demographic data, aggregated financial data like Zip+4 household wealth, and interactive data like social graphs and e-mail addresses. These 50 databases add up to a combined 300Bn records and 200TB of data covering 99% of US adult consumers. All of this is available for every modeling project – some 50,000 attributes applied to every problem. Obviously this has to be integrated in terms of identity matching and in terms of managing data granularity to ensure summary data can be used as well as individual data. This is all done in-house in a highly secure data center in St. Cloud, Minnesota just north of Minneapolis.

One of eBureau’s education clients wanted to predict which leads would result in enrolled students. Over a period of 6 months, this university purchased 537,000 consumer leads which ultimately resulted in 6,000 enrolled students, representing a 1% conversion rate. eBureau found some 120 attributes that were predictive across demographic, property, purchase history, etc.  The average cost-per-enrollment was $3,100 across the whole portfolio but by focusing on the highest scoring segments they were able to reduce this to $2,300 saving them tens of millions of dollars; Classic predictive analytic segmentation.

Depending on sales cycles, it can take several months to know if a lead converts or not, but once the model is built eBureau clients get immediate feedback on the quality of a given lead. This allows eBureau clients to rapidly assess a new lead source, understand quality across a portfolio of lead sources, or simply optimize internal campaigns and creative.

Another example is a company using it to segment inbound leads from ads bought on the spot market (where they don’t control the time to run the ads). Using only the prospect’s phone number, this client used a simple green/yellow/red score to prioritize the incoming calls. When volumes are low, the red (low conversion) ones get handled but when things get busy (because lots of ads are running) only the green and yellow get handled and the green’s (top few segments) get prioritized and routed to the appropriate sales people.

Direct marketers have been doing this for years with direct mail lists. For example, credit card marketers don’t send letters to everyone, they pre-screen to find the best potential audience for the offer. Online display ads can be managed similarly, but using predictive models in real-time to understand who it is worth showing a display ad to. eBureau uses their data to build “look-alike” models for a company’s best customers – online consumers who look (statistically) like your best customers. eBureau protects privacy by placing anonymous cookies allowing advertisers to identify a high-propensity “look-alike” prospect and serve exactly the right display ad in real-time without the advertiser ever knowing who the target is or anything about them. This is a nice example of a predictive model keeping private data private while letting people use that data effectively.

eBureau has also invested in web-based reporting and analysis tools to help their clients understand the impact of models such as score trends by source over time. As with many predictive model and scoring solutions, the education of marketplace and helping new clients with organizational change implications are critical to success.

Personally I think this kind of hosted decisioning/analytics is a great way for many companies to get started with analytics, with using external data sources to enrich their own and to apply analytics in their operational systems.

Here’s my presentation from Predictive Analytics World, reproduced with permission from Predictive Analytics World and Rising Media.Analytic Journeys from Predictive Analytics World