1) The document discusses adopting an effective decision making framework using common methods and processes. It emphasizes capturing analytic insight as an asset and improving collaboration.
2) Decision Model and Notation (DMN) is presented as a common language that can be used to model decisions. DMN provides constructs to define decisions, their requirements, and relationships in diagrams.
3) Examples of applying DMN to decisions around retail conversion rate are shown. Components such as decisions, data sources, and knowledge bases are modeled visually.
2. Introduction
Jim Parnitzke
Big Data Analytics, Enterprise Architecture
Advisor, Expert, Trusted Partner, and Publisher
Hands-on technology executive, trusted partner, advisor, software publisher, and widely recognized
information management and architecture thought leader. Over his career, Jim has served in executive,
technical, publisher (commercial software), and practice management roles across a wide range of industries.
Using analytic insight to solve the toughest problems
Contact:
(c) 904.607.6299
Linked In: http://www.linkedin.com/in/jimparnitzke
Twitter: http://twitter.com/jparnitzke
j.parnitzke@comcast.net
jim.parnitzke@gmail.com
5. Analytics is important…
Making better decisions is more important.
Faster, better outcomes
Precise answers for hard-to-solve problems
Uncovering new growth opportunities
Increasing competitive capability
Improving business results
6. How important?
• Fact:
– over 94% of all Companies have Big Data and Analytics in their top 10
priorities to enable better decision making
– Organizations competing on analytics outperform their peers
8. Decision (Meriam-Webster Dictionary)
1 a : the act or process of deciding
b : a determination arrived at after consideration : conclusion <make a
decision>
2 : a report of a conclusion <a 5-page decision>
3 : promptness and firmness in deciding : determination <acting with decision>
4 a : win; specifically : a victory in boxing decided on points <a unanimous decision>
b : a win or loss officially credited to a pitcher in baseball <has five wins in eight decisions>
9. Organizations make decisions every day…
• Strategic: Few in number, large impact
– Should we acquire this company or exit this market segment?
• Tactical: Management and control, moderate impact
– Re-organize the supply chain
– Change risk management approach
• Operational: Day-to-day decisions
– Improve conversion rates
– Select next best offer for a customer
– Select the terms for a loan
– Which supplier to use
– How to handle this claim; which need to be fast-tracked
11. Marketing analytics drive business decisions daily
Customer Profiling and Segmentation
Up Sell Opportunity Analysis
Real-time Product and Service Recommendations
Try and Buy Usage Analytics
Next Best Offer
Retention Analytics
Digital Marketing and Path to Purchase Analysis
Sell through and Sales Channel Analysis
Lead to Cash
Path to Purchase
Multi-Channel and Attribution Analysis
Customer Lifetime Value
Forecasting (Time Series Analysis)
14. Where is analytic insight captured in your organization?
Is the Intellectual Property managed like any other asset?
Are standard tools and methods used to model, capture,
manage, and share this significant property?
Have you even tested the decisions modeled or executed
through peer review or common diagnostics for defects?
Has your modeling approach evolved at the same speed as
the new technologies and tools used to accelerate the
decision making process?
Is proven practice shared with others?
15. If you answered yes to all five questions
congratulations…
you are exceptional
17. We have used a variety of techniques to accurately describe
the requirements for legacy information systems.
They work pretty well.
Many now realize that current approaches
do not solve the decision-making need that is important to
capturing and managing analytic insight to drive better
outcomes.
For the rest of us…
18. Actionable
Readily understandable by business users
Used by business analysts to create decision requirements
and models
Implemented by technical developers responsible for
automating the decisions in processes
Managed and monitored by stakeholders who own the
results and outcomes of the decisions
The ideal decision framework should be:
19. Better decisions share common values
Actionable. - analysis for analysis sake is ridiculous.
They begin with the right questions. Value is demonstrated in
defining the decision in advance. Learn what data and metrics
are important and make a difference.
Non-trivial. Enough said…
Measurable. Know which measures matter and which don’t.
21. An effective decision making framework
Uses common methods and processes with clear goals and
objectives to manage and measure outcomes
Adopts a common language across business, IT and analytic
communities improving meaningful communication
Captures and manages analytic insight like any other asset
Improves collaboration, increases reuse, shares proven
practice to solve complex problems once
Eases implementation and deployment of decision services
24. Use common methods and processes
Cross Industry Standard Process for Data Mining (CRISP-DM) is a process model that
describes commonly used analytic approaches. It is the leading methodology with 3-4
times as many people using this model as Sample, Explore, Modify, Model and Assess
(SEMMA) developed by the SAS Institute Inc.
25. Design and build independent
decision services using business
rules and advanced analytics
Decision Services
Legacy Systems Websites Business Process Event Correlation Enterprise Applications Mobile
Decision
Service
Business Rules Predictive Analytics
Data Warehouse, Operational Data Stores, Big Data
Create a closed loop between
operations and analytics to
measure results and drive
improvement
Decision Analysis
Identify and model the decisions
that are most important to
operational processes
Decision Discovery
Decision Model Notation
ActionDecision
Actionable goals and objectives
26. Adopt a common language
Decision Model Notation
The OMG Decision Model and Notation standard provides a common
notation; a standardized bridge for the gap between the business decision
design and decision implementation.
The purpose of DMN is to provide the constructs that are needed to model
decisions, so that organizational decision-making can be readily depicted in
diagrams, accurately defined by business analysts. Addresses two different
perspectives by existing modeling standards:
Business process models (e.g. BPMN) can describe the coordination of
decision-making within business processes by defining specific tasks or
activities within which the decision-making takes place.
Decision logic (e.g. PRR, PMML) can define the specific logic used to
make individual decisions, for example as business rules, decision
tables, or executable analytic model
27. Capture Analytic Insight
Where to start
DMN provides a third perspective – the Decision Requirements Diagram
Business process models define tasks within business processes where
decision-making is required to occur. Decision Requirements Diagrams
will define the decisions to be made in those tasks, their
interrelationships, and their requirements for decision logic
Decision logic (business rules) will define the required decisions in
sufficient detail to allow validation and automation.
Taken together, Decision Requirements Diagrams and decision logic can
provide a complete decision model which complements a business process
model by specifying in detail the decision-making carried out in process tasks.
35. Retail Conversion Rate
Retail Conversion Rate
Retail Conversion
Rate Data Flows DRAFT
SIZ
E
FSCM NO DWG NO
RE
V
SCALE 1 : 1 SHEET 1 OF 11
Assemble
Demographic
Profile
Filtered and Validated
Close (Conversion) Rate
Data Set
Assemble
Conversion Rate
Reporting in
context
Assembly Database
ETL
Transformation
Data Warehouse
Actual Sales and Payroll
FPA (acronym)
Store level weather.
Actuals and two-week forecast
Transformed
Weather Forecast
Enterprise
Reporting Platform
Retail Selling Channels
Point of Sales
Order Fulfillment
Installers
Designers
Retail Sales
Transactions
Location, Camera, Count Type,
Date, Time Interval, Count
Report Scheduler
Video Imaging
Video Management System
Video Business Intelligence
Analytic Publisher
Report Generator
Retail Traffic Database
Retail Traffic Counts
Traffic Reporting
Time and Attendance
Kronos
Time and Attendance
Scheduling
Accuweather Pro
WSI Weather Service
Employee actual hours reported at store i on day t
Assemble
Employee
Values
Assemble
Retail Traffic
Values
Customers who entered per period
store , camera, day
Assemble
Sales and
Transaction Counts
for Store for period
Retail Sales
Transactions
Per capita income for store location,
number of like-kind stores within x miles
where store is located,
Intra-day traffic variability for store locationCustomer Profile
Data Warehouse
Conversion Rate
Data Sets
Formatted
Conversion Rate
Reporting
Assemble
Weather Related
Values
Detailed weather forecast information.
Assemble Raw
Video Formats
Raw Video Feeds captured at store i on day t
HDFS
Transactions
Employees
Video
Retail Traffic Weather
Close (Conversion) Rate Data
Captured in Hadoop
Apply Business
Rules and
Assemble
Source Data
Total number of customers who entered store i on day t
Average number of customers who entered per hour store i on day t
Sales volume for store i on day t
Average sales volume per period for store i on day t
Number of customer transactions per period at store i on day t
Average number of transactions per period for store i on day t
Proportion of customers who made a transaction at store i on day t
Value in U.S. dollars of customers' shopping basket at store i on day t
Total number of employee hours reported at store i on day t
Average no. of employee hours per hour reported at store i on day t
Total number of like-kind stores within x miles where store i is located
Daily temperature for store location i
Per capita income for store location i
Average inter-day traffic variability for store location i
Intra-day traffic variability for store location i on day t
Growth in average traffic for store location i in period p
Average conversion rate for store location i in period p-l
Filtered Conversion
Rate Data Set
Per capita income for store location,
number of like-kind stores within x miles
where store is located,
Intra-day traffic variability for store location
Employee forecasted hours reported at store i on day t
Reporter Client
Reporting Results
Conversion Rate Processing
Results Captured in Assembly
Platform
Analytics Repository
Analytic Data Sets
Check for Additional Values:
Seasonality Vectors
Macro-economic Sets
Other data values
Check for Additional Values:
Seasonality Vectors
Macro-economic Sets
Other data values
37. Components
1) Decision Customer Traffic
2) Decision Items per purchase
3) Decision Gross margin
4) Decision Transaction Count
5) Decision Total number of customers who entered store i on day t
6) Decision Average number of customers who entered per hour store i on day t
7) Decision Number of customer transactions at store i on day t
8) Decision Average number of transactions per hour for store i on day t
9) Decision Sales volume for store i on day t
10) Decision Average sales volume per hour for store i on day t
11) Decision Value in U.S. dollars of customers' shopping basket at store i on day t
12) Decision Average inter-day traffic variability for store location i
13) Decision Intra-day traffic variability for store location i on day t
14) Decision Growth in average traffic for store location i in period p
15) Decision Average conversion rate for store location i in period p-l
16) Decision Proportion of customers who made a transaction at store i on day t
17) Decision Daily weather for store location i
18) Decision Total number of employee hours reported at store i on day t
19) Decision Average no. of employee hours per hour reported at store i on day t
20) Decision Conversion Rate
21) Decision Average purchase value
1) Data Source Retail Traffic Count
2) Data Source Point of Sale
3) Data Source Weather Information
4) Data Source Location Time and Attendence
1) Know How Retail Conversion Knowledge
2) Know How Time Series Analysis
3) Know How Time Series Forecast
Decisions
Data Sources
Know How (Knowledge Base)
40. The framework you have seen:
Uses common methods and processes with clear goals and
objectives to manage and measure outcomes
Adopts a common language across business, IT and analytic
communities improving meaningful communication
Captures and manages analytic insight as any other asset
Improves collaboration, increases reuse, and shares proven
practice to solve complex problems once
Eases implementation and deployment of decision services
41. Now you can capture and manage analytic
insight to make better decisions…
Faster, better outcomes
More precise answers
Solve hard problems once
Uncover new growth opportunities
Increase competitive capability
Improve business results
42. Thank You…
Jim Parnitzke
Big Data Analytics, Enterprise Architecture
Advisor, Expert, Trusted Partner, and Publisher
Hands-on technology executive, trusted partner, advisor, software publisher, and widely recognized
information management and architecture thought leader. Over his career, Jim has served in
executive, technical, publisher (commercial software), and practice management roles across a wide
range of industries.
Contact:
(c) 904.607.6299
Linked In: http://www.linkedin.com/in/jimparnitzke
Twitter: http://twitter.com/jparnitzke
j.parnitzke@comcast.net
jim.parnitzke@gmail.com
james.parnitzke@nascentblue.com
If you interested in using analytic insight to solve your toughest problems contact me
below and learn how powerful easy-to-use tools can help across a wide variety of pre-
defined subject areas. Information management, Big Data, Analytics, Master Data
Management, Program and Project Management libraries and tools are available today
to begin delivering fast, quick, and distinctive results.
To get started contact me when ready, look forward to helping us succeed together.
46. Business rules can be expressed using modeling approaches to include:
• Unified Modeling Language
• Business Process Execution Language
• Business Process Modeling Notation
• Decision Model and Notation
• Semantics of Business Vocabulary and Business Rules (SBVR)
47. A simple example…
Gather platform
characteristics profiles
Determine the analytics
operating model in use (can
be more than one)
Gather user profiles and
population counts for each
form of use
Develop platform and tool
signatures
Gather current state
portfolio, interview
stakeholders, and document
findings Refine Critical Analytic
Capabilities as defined
to meet site specific
needs
Weight Critical
Analytic Capability
according to each
operating model in use
Gather data points and
align with the findings
Assemble findings and
complete decision
models for platform
and tooling
optimization
1
2
3
4
5
6
7
8
9
CRISP-DM Method
Nine Step Method
Sample Platform and Tooling Optimization Decision Model
Notes de l'éditeur
Organizations are significantly more likely to outperform their peers.
Organizations achieving competitive advantage with analytics are 220% more likely to be substantially outperform their industry peers
Source:
Survey Chimp, 2015; Big Data Analytics Guide: 2012 http://fm.sap.com/data/UPLOAD/files/SAP_ANALYTICS2012_WEB_ALL_PGS.pdf
IBM IBV/MIT Sloan Management Review Study 2011
Copyright Massachusetts Institute of Technology 2011
Product Recommendation Engine
Recommend a product item or category based on historical or recent purchases and interactions
Recommend (at the appropriate time) the products that customers would repurchase (air filters, light bulbs, etc.)
Recommend list of products for a customer to purchase or add to their account
Recommend local store-level items that are on clearance to drive trips
Product Affinity Targeting - Determine the best group of customers to include in product-focused email or direct mail campaigns
Simple Product Recommendation
Recommend a product item or category based on historical or recent purchases and interactions
Recommend (at the appropriate time) the products that customers would repurchase (air filters, light bulbs, etc.)
Recommend list of products for a customer to purchase or add to their account
Recommend local store-level items that are on clearance to drive trips
Product Affinity Targeting - Determine the best group of customers to include in product-focused email or direct mail campaigns