Decision Services for SAI

Decision Services A pattern for business rules in Service Oriented Systems and Architectures James Taylor Principal Smart (enough) Systems LLC

Agenda

Smarter Systems

Decision Services Defined

Building Decision Services

Decision Services and SOA

Decision Services and BPM

Dumb Systems Are Everywhere… Report but don’t learn Built to last, not to change Wait rather than act CRUD

Why Smarter Systems? Decision-Making Well-Defined Increasingly Complex Timeliness Days Real-time Objectives Local and clear Complex Trade-Offs Regulations National and simple Complex and global Changes to Strategy Every 3-5 Years Constant Operational Volume Low High

So What IS A Smarter System

Operational

Real-Time

Rapidly evolving - agile

Learning

Demonstrably Compliant

Cost-Effective

Business-Driven

Smarter Systems Make More Decisions People Not Made Embedded People Embedded Not Made New Before After Larger boxes represent more decisions, by volume

Different kinds of decisions Low High High ECONOMIC IMPACT OF INDIVIDUAL DECISION Low DECISION VOLUME High-value, low-volume decisions Medium-value, medium-volume decisions Low-value, high-volume decisions

Decisions, Decisions, Decisions

Agenda

Smarter Systems

Decision Services – A Pattern

Automated decision systems have been a common element of Financial Services application portfolios for some time

With the advent of SOA and component architectures, the idea of a separate Decision Service has become widely accepted

As the cost of technology has fallen and awareness increased, the approach has been validated

Applications have evolved Data Process Logic User Interface Browser BPM Services

Evolution Completed

A self-contained, callable service with a view of all the conditions and actions that need to be considered to make an operational business decision.

Decision Services

A service that answers a business question for other services and processes.

What is a decision service? Services

Problem

Some business services implement decisions

These services must follow policies and regulations

These services must be easy to change

These services must be driven “by the numbers”

These services simplify business processes

These services often replace high-change elements of legacy applications

Many Forces At Work

Agenda

Smarter Systems

The Solution

ENTERPRISE DECISION MANAGEMENT is an approach for automating and improving high-volume operational decisions.

Focusing on operational decisions,

it develops decision services

using business rules to automate those decisions,

adds analytic insight to these services using predictive analytics

and allows for the ongoing improvement of decision-making through adaptive control and optimization .

Applicability

Decisions where large numbers of policies or regulations apply

Decisions with policies and regulations that change a lot

Decisions that require business domain knowledge to understand

Decisions that are complex or have complex interactions

Decisions that the business insists on owning

Decisions requiring analytic insight

Decisions that cut across events and processes

Consequences Isolation of Decisions Control and Agility in Business Processes Business Control of Decisions Integration Points for Advanced Analytics Externalization from Legacy Applications

Context

Delivering Decision Management

Business Rules Are Everywhere Experienced Personnel Regulations Policy Manuals Legacy Systems Historical Data Managed Business Rules

Manageable Business Rules Smart (Enough) Systems, Prentice Hall June 2007. Fig 4.3 If customer is GoldCustomer and Home_Equity_Loan_Value is more than $100,000 then college_loan_discount = 0.5% If member has greater than 3 prescriptions and prescription’s renewal_date is less than 30 days in the future then set reminder=“email” If patient’s age is less than 18 and member’s coverage is “standard” and member’s number_of_claims does not exceed 4 then set patient’s coverage to “standard”

Decision Flow

Smart (Enough) Systems, Prentice Hall June 2007. Fig 6.2

score >= 3 score < 3 true true true true approval is true approval is false Loan Decision Reject Application Generate Report Account Coordination Notify Customer Service Billing Adjustment Reject Application Price Structuring Booking

A list of rules representing a rule set

AdjustIncome RuleSet (newApplicant) Name: Six months or less on the job if newApplicant’s monthsInCurrentJob is less than or equal to 6 then decrement newApplicant’s income by 5600 Name: About 1 year if newApplication’s monthsInCurrentJob is between 6 and 13 then increment newApplicant’s income by 1000 Name: Greater than 1 year if newApplicant’s monthsInCurrentJob is greater than 12 then increment newApplicant’s income by 2500 Name: Computer Total Income if newApplicant is married and newApplicant’s spousalIncome is greater than 0 then newApplicant’s totalIncome is equal to newApplicant’s income plus newApplicant’s spousalIncome else newApplicant’s totalIncome is equal to newApplicant’s income Name: Minimum Income if newApplicant’s totalIncome is less than 40000 then newApplicant is not eligible

A selection of decision table layouts

I n c ome C a r d T ype C r edit Limit 2,500 P oor 25,000 - 34,999 3,000 G ood 25,000 - 34,999 4,000 E x c elle n t 25,000 - 34,999 3,000 P oor 35,000 - 44,999 3,500 G ood 35,000 - 44,999 I n c ome C a r d T ype S tude n t B r on z e 25,001 - 35,000 25,001 - 35,000 S tude n t G old 25,001 - 35,000 35,001 - 45,000 35,001 - 45,000 35,001 - 45,000 > 45,000 > 45,000 > 45,000 S tude n t P l a tinum S tude n t B r on z e S tude n t G old S tude n t G old S tude n t B r on z e S tude n t P l a tinum S tude n t P l a tinum C r edit Limit 2,500 3,000 4,000 3,000 3,500 4,500 4,200 4,700 5,200 S tude n t B r on z e 2 - Axis G r id 1 - Axis V e r tical 1 - Axis Ho r i z o n tal C a r d T ype C r edit R a ting Home O wner? S tude n t B r on z e P oor false C r edit Limit S tude n t P l a tinum E x c elle n t true C r edit Limit 4,000 4,500 5,200 S tude n t G old G ood true C r edit Limit 3,000 3,500 4,700 I n c ome 25,001 - 35,000 35,001 - 45,000 >45,000 2,500 3,000 4,200

A decision tree

IsEmpl o y ed C r editEstim a t e Y ea r ly I n c ome Y ea r ly I n c ome L oan T ype R e f e r r ed poor othe r wise fair e x c elle n t true Y ea r ly I n c ome L oan Limit 20,000 L oan Limit 10,000 L oan Limit 5,000 L oan Limit 3,000 L oan Limit 2,000 L oan Limit 1,000 L oan Limit 1,000 L oan Limit 500 L oan Limit 1,500 < 25,000 25,00 - 75,000 > 75,000 > 90,000 < 100,000 < 40,000 40,000 - 90,000 50,000 - 100,000 < 50,000

Help the Business to Manage Decisions So you business-types want to be able to change your business rules? I want to relax my underwriting policy I want to be able to promote a new product combination I need to add the new regulations No…

Which Would Your Users Understand? Smart (Enough) Systems, Prentice Hall June 2007. Fig 6.8 public class Application { private Customer customers[]; private Customer goldCustomers[]; ... public void checkOrder() { for (int i = 0; i < numCustomers; i++) { Customer aCustomer = customers[i]; if (aCustomer.checkIfGold()) { numGoldCustomers++; goldCustomers[numGoldCustomers] = aCustomer; if (aCustomer.getCurrentOrder().getAmount() > 100000) aCustomer.setSpecialDiscount (0.05); } } } If customer is GoldCustomer and Home_Equity_Loan_Value is more than $100,000 then college_loan_discount = 0.5% College Loan Discounts Current Discount = Eligibility Gold Customer and Home Equity Loan more than $100,000 %

Rule templates to make maintenance simple Smart (Enough) Systems, Prentice Hall June 2007. Fig 6.7 Rules R eposi t o r y Rule: M issing I t ems: and or S t r eet A dd r ess // I n c omple t e add r ess // I f A pplica n t ’ s zip is missing or A pplica n t ’ s ci t y is missing and A pplic a tio n ’ s st a t e is missing or A pplica n t ’ s st r eet A dd r ess is missing then set c omple t eness of A pplic a tion t o F ALSE // I nsufficie n t Name // I f A pplica n t ’ s LastName is missing then set c omple t eness of A pplic a tion t o F ALSE // A dd r ess does not e xist // I f v e r ifyUSPS A dd r essCheck ( A pplica n t ’ s add r ess) is F ALSE then set eli g ibili t y of A pplic a tion t o F ALSE // T oo y oung and poor // I f A pplica n t ’ s age is less than 30 and A pplica n t ’ s I n c ome is less than $100,000 or A pplica n t ’ s age is less than 25 and A pplica n t ’ s in c ome is less than $150,000 then set eli g ibili t y of A pplic a tion t o F ALSE C onfi r m Eli g ibili t y Rules A pplic a tion . c omple t eness .eli g ibili t y . A pplica n t .lastname . . . . P oli c y R equest . . . D a ta C omple t eness T empl a t e K noc k - out T empl a t e Rule: and or A ge C i t y S t a t e Zip Last Name M o r e than < v alue> L ess than < v alue> B e t w een <a> and <b> I n c ome G ender R isk S c o r e A dd r ess V alidi t y

Decisions remain in sync with market changes Smart (Enough) Systems, Prentice Hall June 2007. Fig 2.11 Business users Conventional Approach Enterprise Decision Management Programmers Frequent code changes CRM System Other Systems Frequent policy changes Infrequent code changes Decision Service CRM System Other Systems Programmers Policy Changes Business users so r t ed=1: f or I y = I–I= y > 11; y – I t p r i n t ( y – Id ,y) : if [1a r r a y I y] < 1a r r a y I y -1]) t holder - [a r r a y[ y -1] | Ia r r a yl y -1]) = 1 [y]) 1 a r r a y = holder so r t ed=1: f or I y = I–I= y > 11; y – I t p r i n t ( y – Id ,y) : if [1a r r a y I y] < 1a r r a y I y -1]) t holder - [a r r a y[ y -1] | Ia r r a yl y -1]) = 1 [y]) 1 a r r a y = holder

Analytics <> Graphs

Even graphics require interpretation

Not everyone can see the patterns

And code does not “see” at all

Data Mining - Improve Rules * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Low-moderate income, young High Income High income, low-moderate education Moderate-high education low-moderate income High Moderate education, low income, middle-aged Low education, low income Education Age High

Predictive Analytics – Add Insight 10 20 30 40 Member renews Member fails to renew

Impact May Take Time to Play Out

Use Adaptive Control to Continuously Improve

The Evolution Of An Interaction

Automate Decision

Apply rules

Segment customers

Predict risk, value

Optimize decision

Web http://www.f Email Call Center Mobile

Building Decision Services Smart (Enough) Systems, Prentice Hall June 2007. Fig 5.1 Production Process Enterprise IT Infrastructure Data Warehouse Operational Data Store Business Rules Analytics Decision Service Adaptive Control Events

Analytic Modeling Analytic Modeling Smart (Enough) Systems, Prentice Hall June 2007. Fig 5.6 Data Warehouse Model Repository Data Mining and Analysis Data Preparation Data Requirements Determine Outcome Model Deployment Model Validation Model Tuning Model Development Predictive Model Rules

Rule Development Smart (Enough) Systems, Prentice Hall June 2007. Fig 6.9 Business Rules Rules Repository Rules Deployment Rules Validation and Testing Reporting Dashboards Rules Development Rules Maintenance Model Repository Predictive Model Rules Rules Rules

Adaptive Control Smart (Enough) Systems, Prentice Hall June 2007. Fig 7.9 Adaptive Control Rules Repository Experimental Design Champion/ Challenger Setup Simulation Optimization Model Design Model Repository Decision Service Data Warehouse Model Repository Decision Analysis Predictive Model Rules

Business Rules Management Technology Smart (Enough) Systems, Prentice Hall June 2007. Fig 6.6 Design Tools Rule Management Applications Business Application Decision Service Rule Engine Customers Employees Partners Suppliers IT Staff Business Experts Data Sources Rule Repository

Business Rules and Decision Services

Adaptive Control Inside a Decision Service

Deployed Decision Services Smart (Enough) Systems, Prentice Hall June 2007. Fig 4.5 ERP CRM Billing SCM f (s) Business Intelligence Analytic Development Rules Management Decision Analysis Process Data, Rules and Analytics to Determine Best Decision Decision Service Data Warehouse Request Decision Decision Channels Call Center Web Email Telemarketing Direct Mail Store/Branch Kiosk/ATM Field

Agenda

Smarter Systems

8 Principles of Service-Orientation

Standardized service contract

Service Loose Coupling

Service Abstraction

Service Reusability

Service Autonomy

Service Statelessness

Service Discoverability

Service Composability

7 Goals of Service-Orientation Goal Decision Management Impact Increased intrinsic interoperability Increased federation Increase business and technology domain alignment Business and IT share an understanding of business logic Increased vendor diversification options Increased ROI Increased Organizational agility Decision Services are more transparent and agile Decreased IT burden Empowering the business to maintain decision services

4 Characteristics of SOA

Business-Driven

Vendor-Neutral

Enterprise-Centric

Composition-Centric

Benefits Of Decision Services

“ Built to change not built to last”

In the real world, nothing is set in stone

What do you do if a service must change a lot?

Managing “change-time”

Legacy Modernization

Externalize decisions hidden in legacy applications

A bridge between your mainframe and your SOA

Differentiation

Buy services based on best practices, standards

Buy services that don’t differentiate you

Build Decision Services for competitive differentiation

A Central Credit Decision Service

Business Rules A nalytics I nput EL C alcul a tion D ecision S e r vi c e E x t e r nal D a ta I n t e r nal D a ta S ales D a ta L oan D a ta R a ting R esults Business Rules Black List O utput S ales S y s t ems I n t e r fa c e L a y er PD C alcul a tion

Decision Service and Message Bus

Business P r o c ess M anageme n t / W o r kfl o w M anageme n t M essage Bus A pplic a tion S e r v er P r odu c tion S y s t ems Business Rule R eposi t o r y A c ti v e D ecision En g ine T r ansa c tion w aiting T r ansa c tion p r o c essing T r ansa c tion p r o c essed D ecision S e r v er Subsc r ibe I n t e r fa c e P ublish I n t e r fa c e D ecision S e r vi c e A

Decision Services in Mortgage Lending

Agenda

Smarter Systems

An Underwriting Example  Process Tasks  Decision Tasks

May include customer visits, collection of forms, etc.

Triggers more data gathering process steps

Triggers rejection process

Calculate risk and determine price, may trigger exception handling

Risk and Benefits

Process without Decisions

Decisions an afterthought

Decisions buried in process

Process becomes complex

Inconsistency of rules likely

Decisions only evolve with process

Hard to share decisions

Process with Decisions

Decisions first class object

Decisions are linked but not buried

New process is simplified

Independent process & decision changes

Decisions (and Decision Services) are reusable

Decision logic <> Process logic

Process Management

Standardizes processes How should a process be carried out?

Facilitates collaboration and workflow

Process automation around decision-making

Workflow definition and management

Integration broker

Decision Management

Standardizes decision-making What should the decision be based on?

Facilitates decision automation and maintenance

Replaces manual decision-making within processes

Business rules definition and management

Decision broker

Decisions and Processes Complementary but not the same

Decision Services in BPM

Processes become complex when decision making is mismanaged

Explicit decision services simplify processes dramatically

All Kinds of BPM Need Decisions Sense and Respond Systems People-Centric Automation Transaction-Centric Automation ESB CEP BAM Process Analytics Process Warehouse Decision Services Investigate Exceptions Automation After IDC Group “A View of BPM in 2007 and beyond”, Maureen Fleming

Wrap Up

Smarter Systems are Needed

Systems must make more decisions

In an SOA world this means Decision Services

Decision Services are a pattern

Well proven

Built on existing technology

Compatible with Rules Centralization

Ready for analytics

Known Uses

Recommendations

Next Best Action

Underwriting

Fraud Detection

Eligibility

Scheduling

Accounting Allocation

Billing

Dynamic Pricing

…

Action Plan

Identify your decisions

Decisions that matter to customers

Hidden decisions

Transactional decisions

Consider

Who takes them now

What drives changes in them

What the context is for them

Assess

Change Readiness

Technology adoption

Organizational change

Adopt Business Rules

Approach and technology

Management and governance

Change the relationship between business and IT

Investigate Data Mining and Predictive Analytics

Data Mining for rules

Predictive reporting

Executable analytics

Build Adaptive Control into your applications

An overview of traversing the steps in EDM Smart (Enough) Systems, Prentice Hall June 2007. Fig 9.1 Phase 1 - Piecemeal Phase 3 - Expansion Readiness Assessment Phase 2 – Local Decision Management Rules and Analytics Champion Challenger Steady State EDM Enterprise Management Analytic Value Chain Optimization and What If Enterprise Backbone Broaden Analytic Base Improved Foundation Foundation First Rules Project First Analytic Project Manage Scenarios Overlapping and Adjacent Projects

Smart (Enough) Systems – The Book

How key business trends impact the decision-making process

Why organizations need systems smart enough to cope with these trends

How decision automation can make their systems smart enough

How to translate decisions into a corporate asset and competitive advantage

The ROI and business impact of better decisions and smarter systems

The core concepts and technologies needed and how they work together

The book is full of insightful examples of problems solved by applying Enterprise Decision Management across various industries and outlines a practical and incremental method for implementing the technology.

Thank You James Taylor [email_address] http://www.smartenoughsystems.com Blog : www.smartenoughsystems.com/wp