BERLIN

LONDON
MADRID
PARIS
TURIN
WARSAW

Data Modeling (System Analysis & Design)

MSc Big Data & Business Analytics

Session 2: Process modeling with DFDs

Professors
Louis-David Benyayer / Ghali Bensouda

Session 2 2-1
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs

E. Real use-case

2
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs

E. Real use-case

3
CLASS RULES
✓3 hours sessions with 15 minutes break.
✓Intervention method : someone in the classroom to ask the
questions for the ones in remote.
✓Interactive & practical course as much as possible:
exercises and real-world use case.
✓Ask questions or make comments whenever you want
✓Evaluation method: an open-book exam

Session 2 2-4
LEARNING OBJECTIVES (1/2)
✓Understand the process modeling by studying examples
of data flow diagrams (DFDs).
✓Draw data flow diagrams following specific rules and
guidelines that lead to accurate and well-structured
process models.
✓Decompose data flow diagrams into lower-level
diagrams.
✓Balance higher-level and lower-level data flow
diagrams.

Session 2 2-5
5
LEARNING OBJECTIVES (2/2)
✓Use data flow diagrams as a tool to support the analysis
of information systems.
✓Discuss a real data modeling use case for an
international car manufacturer.

Session 2 2-6
PROCESS MODELING IS DONE DURING
THE ANALYSIS PHASE

FIGURE 2-1
Systems development life cycle with the analysis phase highlighted

Session 2 2-7
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs

E. Real use-case

8
 PROCESS MODELING IS ONE OF THE THREE
COMPLEMENTARY VIEWS OF A SYSTEM

Process Logic
modeling modeling
(1) (2)

Data
modeling
(3)
(1) Identify processes, data flows & data stores
(2) Detail the logic inside the processes
(3) Structure & relationships within the tables

Session 2 2-9
PROCESS MODELING DEFINITION
Graphically represent the processes that capture, manipulate,
store, and distribute data between a system and its environment
and among system components.

Utilize information gathered during requirements determination.

Model processes and data structures.

Several different tools have been developed for process
modeling. We will focus on DFD as it is the most frequently used
today for process modeling.

To go further, please check this video on UML Class diagram:

Session 2 https://www.youtube.com/watch?v=WI0oyCeon2A&t=287s&ab_channel=SmartDraw 2-10
DFD DEFINITION
A Data Flow Diagram (DFD) is a picture of the movement of
data between external entities and the processes and data
stores within a system.

FIGURE 2-2
Example of a Data Flow Diagram

Session 2 2-11
PROCESS MODELING DELIVERABLES
Context data flow diagram (DFD)
 Delimits the scope of system
DFDs of current physical system
 Specify people & technology used in each processes
DFDs of current logical system
 Enable analysts to understand current system
DFDs of new logical system
 Show what data processing functions the new systems will perform
Thorough descriptions of each DFD component
 Centralized documentation in a shared repository

Session 2 2-12
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs

E. Real use-case

13
FOUR DIFFERENT TYPES OF DFDS
Current Physical
 Process labels identify technology (people or systems) used to process the
data.
 Data flows and data stores identify actual name of the physical media.
Current Logical
 Physical aspects of system are removed as much as possible.
 Current system is reduced to data and processes that transform them.
New Logical
 Includes additional functions
 Obsolete functions are removed.
 Inefficient data flows are reorganized.
New Physical
 Represents the physical implementation of the new system.

Session 2 2-14
SYMBOLS USED IN DFD - REPRESENTATION

FIGURE 2-3
DeMarco and Yourdon
VS Gane and Sarson DFD symbol
sets

Session 2 2-15
SYMBOLS USED IN DFD - DEFINITION

Process: work or actions performed on data (inside the

system)
Data store: data at rest (inside the system)
Source/sink: external entity that is the origin or
destination of data (outside the system)
Data flow: arrows depicting movement of data

Session 2 2-16
CONTEXT DIAGRAM - DEFINITION

Context diagram is an overview of an organizational system

that shows:
 the system boundaries.
 external entities that interact with the system.
 major information flows between the entities and the system.

Note: only one process symbol, and no data stores shown

Session 2 2-17
CONTEXT DIAGRAM - REPRESENTATION

FIGURE 2-4
Context diagram of Hoosier Burger’s food-ordering system

Session 2 2-18
LEVEL-0 DIAGRAM - DEFINITION

Level-0 diagram is a data flow diagram that represents a

system’s major processes, data flows, and data stores at a
high level of detail.
 Processes are labeled 1.0, 2.0, etc. These will be decomposed into more
lower-level DFDs.

Session 2 2-19
LEVEL-0 DIAGRAM - REPRESENTATION

FIGURE 2-5
Level-0 DFD of Hoosier Burger’s
food-ordering system

Session 2 2-20
VIDEO RECAP OF WHAT A DFD IS

https://www.youtube.com/watch?v=6VGTvgaJllM&ab_channel=SmartDraw

Session 2 2-21
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs
1. DFD rules
2. Decomposition & Balancing
3. Guidelines for drawing DFDs
E. Real use-case

22
DATA FLOW DIAGRAMMING RULES

There is a set of rules that you must follow when drawing

DFDs.
There are two DFD guidelines that apply:
 The inputs to a process are different from the outputs of that process.
 Processes purpose is to transform inputs into outputs.
 Objects on a DFD have unique names.
 Every process has a unique name.

Session 2 2-23
DATA FLOW DIAGRAMMING RULES

FIGURE 2-6
Rules Governing Data
Flow diagramming

Session 2 2-24
DATA FLOW DIAGRAMMING RULES

FIGURE 2-7
Rules Governing Data
Flow diagramming

Session 2 2-25
DATA FLOW DIAGRAMMING RULES

FIGURE 2-8
Incorrect and correct ways to draw DFDs

Session 2 2-26
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs
1. DFD rules
2. Decomposition & Balancing
3. Guidelines for drawing DFDs
E. Real use-case

27
DECOMPOSITION OF DFDS - DEFINITION

Functional decomposition is an iterative process of

breaking a system description down into finer and finer
detail.
 Creates a set of charts in which one process on a given chart is
explained in greater detail on another chart.
 Continues until no subprocess can logically be broken down any
further.

Session 2 2-28
DECOMPOSITION OF DFDS - EXAMPLES
Level-1 diagram results from decomposition of Level-0
diagram.
Level-n diagram is a DFD diagram that is the result of n
nested decompositions from a process on a level-0 diagram.

Primitive DFD is the lowest level of a DFD.

Session 2 2-29
 LEVEL-1 DFD

Level-1 DFD shows

the sub-processes of
one of the processes
in the Level-0 DFD.

Processes are labeled 4.1, 4.2, etc. This is a Level-1 DFD

These can be further decomposed in for Process 4.0.
more lower-level DFDs if necessary.

FIGURE 2-9
Level-1 diagram showing the decomposition of Process 4.0 from
the level-0 diagram for Hoosier Burger’s food-ordering system

Session 2 2-30
LEVEL-N DFD
FIGURE 2-10
Level-2 diagram showing the decomposition of Process 4.3 from the
level-1 diagram for Process 4.0 for Hoosier Burger’s food-ordering
system Level-n DFD shows
the sub-processes
of one of the
processes in the
Level n-1 DFD.

This is a Level-2
DFD for Process
Processes are labeled 4.3.1, 4.3.2, etc. If this is the 4.3.
lowest level of the hierarchy, it is called a primitive DFD.

Session 2 2-31
BALANCING DFDS - DEFINITION
Balancing is based on a conservation principles. It consists
on conserving inputs and outputs to a data flow diagram
process when that process is decomposed to a lower level.

Therefore, balanced DFD means:

 Number of inputs to lower level DFD equals number of
inputs to associated process of higher-level DFD
 Number of outputs to lower level DFD equals number of
outputs to associated process of higher-level DFD

Session 2 2-32
BALANCING DFDS - EXAMPLE
(a) Context diagram

1 input
This is
1 output unbalanced
because the
(b) Level-0 diagram
process of the
context
diagram has
2 inputs only one input
but the Level-0
1 output
diagram has
two inputs.

FIGURE 2-11 An unbalanced

set of data flow diagrams
Session 2 2-33
BALANCING DFDS - DATA FLOW SPLITTING

Data flow splitting is when a composite data flow at a

higher level is split and different parts go to different
processes in the lower level DFD.
The DFD remains balanced because the same data is
involved, but split into two parts.

Session 2 2-34
BALANCING DFDS - DATA FLOW SPLITTING
EXAMPLE
(a) Composite data flow

(b) Disaggregated data flows

FIGURE 2-12
Example of data flow splitting

Session 2 2-35
BALANCING DFDS - MORE ADVANCED DFD
RULES

FIGURE 2-13
Advanced rules governing Data Flow Diagramming

Session 2 2-36
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs
1. DFD rules
2. Decomposition & Balancing
3. Guidelines for drawing DFDs
E. Real use-case

37
GUIDELINES FOR DRAWING DFDS
Completeness
 DFD must include all components necessary for system.
 Each component must be fully described in the project dictionary.

Consistency
 The extent to which information contained on one level of a set of nested
DFDs is also included on other levels.

Timing
 Time is not represented well on DFDs.
 Best to draw DFDs as if the system has never started and will never stop.

Session 2 2-38
GUIDELINES FOR DRAWING DFDS
Iterative Development
 Analyst should expect to redraw diagram several times before reaching the
closest approximation to the system being modeled.

Primitive DFDs
 Decision has to be made when to stop decomposition (CF. Slides 40-45)

Session 2 2-39
 RULES FOR STOPPING DECOMPOSITION
• When each process has been reduced to a single
decision, calculation or database operation

• When each data store represents data about a single

entity

• When the system user does not care to see any more
detail

Session 2 2-40
 RULES FOR STOPPING DECOMPOSITION
• When every data flow does not need to be split further
to show that data are handled in various ways

• When you believe that you have shown each business

form or transaction, online display and report as a single
data flow

• When you believe that there is a separate process for

each choice on all lowest-level menu options

Session 2 2-41
USING DFDS AS ANALYSIS TOOLS

Gap Analysis is the process of discovering discrepancies

between two or more sets of data flow diagrams or
discrepancies within a single DFD.
Inefficiencies in a system can often be identified through
DFDs.

Session 2 2-42
A. Introduction

B. Process modeling

C. DFD concepts

D. Developing DFDs
1. DFD rules
2. Decomposition & Balancing
3. Guidelines for drawing DFDs
E. Real use-case

50
 REAL USE-CASE - DESIGNING A SYSTEM
FOR AN INTERNATIONAL CAR PRODUCER
VALUE STATEMENT
Calculate the real contribution of digital leads to new car sales:

• Ensure an industrial, standard & controlled calculation. The first aim of ESCP Reconciliation system is a standardized calculation across markets &
brands of the sales-on-lead ratio KPI.

• Provide end-users with a simple and user-friendly tool to perform regular and bespoke analysis.

• Provide the data in tables the local BI teams can reprocess by themselves

TARGET USERS NEEDS PRODUCT VALUE

• Externalized data matching • Internalization of data

• Data visualization tool with
• Senior executives : poor understanding of matching and analysis (cost
graphs & KPIs per brand
(Corporate) Digital contribution to sales saving)
per country
• Digital performance • Sales on Leads calculation • Automatic, comparable &
• Matching capacity
manager (Corporate) unreliable and more accurate reportings
uncomparable
• Automatic ingestion &
• Digital & CRM manager in • Digital leads conversion
processing into the
regions & countries • CRM/Lead not handled
Datalake
homogeneously • Campaign efficiency

Session 2 2-51
51
 LEADS DEFINITION
▪ A lead is a mean to capture the willingness from a prospect / customer that :
o expresses an interest in our cars or services
o requests to go furtherregarding offers on vehicles, after-sales, services

▪ The lead enables the Group to follow the prospect / customer demand
o all along the Group sales & distributionnetwork
o on all steps of the sales funnel, from the lead capture until the sales

▪ A contact generated from a marketing campaign is not a lead

o the Campaign contact becomes a lead when the customer has expressed his interest on the offer

▪ Lead treatment has to be compliant to data privacy and to respect commercial

agreements collected at any touch point of our distribution channel

Session 2 2-52
 PERSONA OF KEY USER 1 – SENIOR
EXECUTIVE (CORPORATE)
Behaviors
• Define and own the strategy,
and responsible of the end
performance
• Animation of their scope
• Occasional reporting
• Comparison between countries
& regions
Session 2
Needs and pain points Goals
• Strategic necessity to internalize • To perform data based decisions
competencies and to build value on
corporate data • To increase countries efficiency &
better business performance
• Compare real performance vs
considered reference (Y-1, Y Target)
• Have a country/region-based
summary of performance
• Have a broader and clearer
understanding of where to focus:
• key steps of the funnel
• key countries
• best resources allocation
2-53
53
 PERSONA OF KEY USER 2 – DIGITAL OR
CRM MANAGER (COUNTRY)

Behaviors Needs and pain points Goals

• Frequent / Daily usage for
leads analysis, and
campaign performance
analysis
• Provides raw data to build
corporate monthly reporting
• Own the budget
• Data should be reliable and it’s not
always the case
• Focus less on data management
and more on value added analysis
• Have a simple tool to perform deep
and custom analysis based on my
local KPIs
• Provide in a simple way, a reliable
reporting
• Better perform business
performance analysis and
manage perimeter

• Dependent on a costly agency to

perform my job

• Data update must be daily, especially

for campaign management

Session 2 2-54
54
THE ESCP-R SYSTEM VALUE IS BASED ON
THE RELEVANT PROCESSING & MATCHING

COLLECT PROCESS MATCH DISPLAY

X-Leads
Leads

Clean Deduplicate ESCP

Leads Leads R-System:
Local KPIs
and graphs

X-Sales Clean MATCH

Sales

Deduplicate Other
Sales Sales reports
Local

External input Data useable Conversion Meaningful data

data for matching calculation to analyze

Session 2 2-55
 THE ESCP RECONCILIATION SYSTEM
ANSWER TO 3 LEVEL BUSINESS ISSUES
Corporate Teams
▪ Performance monitoring team stop time-consuming
reporting using R-Solve directly
▪ Corporate digital teams measure their own
performance via R-Solve

Countries marketing teams/ agencies Compare to top countries

▪ Understand the performance of campaigns
▪ Monitor performance trend to reach their annual Refer to a single KPI - source of truth
targets
▪ Run focused analyses to adjust their strategy and
prioritize the best digital features

Retarget their segments efficiently Local sales department / network

▪ Monitor dealers’ performance (leads to sales),
Prioritize high converting channels using the Dealer view
▪ Set targets & incentives to their dealers, based on
their conversion performance
▪ Re-allocate leads to the best dealers

Animate the dealer network efficiently

Session 2 2-56
56
 DFD FOR THE ESCP-R SYSTEM

• Who are the sources/sinks of the system?

• What are the inputs and outputs of the system?

• How the data is processed?

• What are the tables we need to create to store the data?

Session 2 2-57
57
CONTEXT DIAGRAM OF ESCP-R SYSTEM

FIGURE 2-14
Context Diagram of ESCP-R System

Session 2 2-58
58
LEVEL-0 DIAGRAM OF ESCP-R SYSTEM

FIGURE 2-15
Level-0 Diagram of ESCP-R System

Session 2 2-59
LEVEL-1 DIAGRAM OF ESCP-R SYSTEM

FIGURE 2-16
Level-1 Diagram of Process 1.0

Session 2 2-60
SUMMARY
In this chapter you learned how to:
✓Understand process modeling via data flow diagrams (DFDs).
✓Draw data flow diagrams of well-structured process models.
✓Decompose data flow diagrams into lower-level diagrams.
✓Balance high-level and low-level data flow diagrams.
✓Use data flow diagrams for analyzing information systems.

Session 2 2-61