Project Estimation

1
Project Estimation

• SEG 3100Software Development for Large Scale
  Systems

2
(No Transcript)
3
Why Project Estimation

• Controlling software development costs
• - Costs exceeds budget in 90 of the cases
• Parkinsons Law
• Controlling software development time
• - 30 of Projects are never completed or are
  cancelled Standish Report
• Better project management ? Tradeoffs between
  Time, Costs, Contents and Quality
• Provides defensible figures at the time of
  submitting/evaluating bids and contracts

4
What is to be Estimated

• Total size of the project (in KLOCs)
• Effort involved (Person-Months)
• Typical duration of the project (Months)
• Optimum number of people
• Distribution of effort and time in the different
  stages of the project
• Expected Errors in the product, i.e. Quality

5
Estimation Process Diagram

• Size
• Count of Use Cases
• Function Points
• Design Points
• Lines of Code

Effort
Development Time
Estimation Process
Number of Personnel

• Factors
• Design Complexity
• Technical
• Environmental
• Personnel

Errors
6
What is needed to do Estimation

• High Level Design
• Requirements Analysis, e.g. Use Cases ? UCP
• High Level Design
• For DP applications Inputs, Outputs, Files..
• For OO applications UML Diagrams
• Nature of the Project
• Product Complexity
• Process Constraints
• Standards Requirements
• Technology Factors
• Personnel Factors Skills

7
Estimation Process

• Step1 Collect Information Size and other
  factors
• Step2 Estimate Lines of the Code
• Step3 Compute Effort
• Step4 Compute Development Time
• Step5 Compute Number of Personnel
• Step6 Distribution of Effort
• Step7 Estimation of Errors

8
Estimation Process strategy

• Step1 Size Calculations
• For general and DP designs use
• Function Points Method ? FP
• For OO based designs
• Count Number of Classes ? FP
• Step2 Estimation of Lines of the Code (LOC)
• Convert FP into LOC
• Direct estimation Count Number of Methods ? LOC
• Step3 Compute Effort
• Apply COCOMO (Constructive Cost Modeling)
  formulas.
• For Use Case Point Method bypass Steps 1 2.
  Count Actors and Use Cases ? Development Time
• Step4 Compute Development Time
• Apply COCOCMO Formulas

9
Methodologies

• A. Use Case Points
• B. Function Points
• C. Object Oriented
• 1. Class base design
• 2. Rational Rose Real Time Design
• D. Miscellaneous

10
Design
Classes2 Function Points Capsulesx Function
Points
Unadjusted Function Point (UFP table)
DI?ratings of selected factors
14 TCF0.650.01?(Factor)j
1 MinTCF0.65 MaxTCF1.35
Modify FPUFPTCF
LOC18.25Num of Methods (modify for RRRT)
TCF
 bKLOC? (LOCs all Classes, Caps)/1000
AFPFP1.25
Error Table Y
Compute Errors AFPY
KLOCMaxaKLOC, bKLOC
uEffort UCP Method
cEffort Person Months, PMA(KLOCB)
Effort, PMMaxcEffort, uEffort
Result
Adjusted PM APM(total EAF)PM
EAFProduct of selected factor
Effort Factors1-15
Development Time TDEVC(APMD)
Number of personnel NPAPM/TDEV
11
A. Use Case Points Karner 1993

• UCPs is and Effort Estimation technique. It
  bypassed the size estimation stage. The number
  and type of Use cases as well as the actors are
  considered.

12
A. Use Case Point Example

• Use Case Name Place Order
• Short description
• The customer provides address information and a
  list of product codes.
• The system confirms the order.
• Basic flow of events
• Customer enters name and address
• Customer enters product codes for items he wishes
  to order
• The system will supply a product description and
  price for each item
• The system will keep a running total of items
  ordered as they are entered
• The customer enters credit card information
• The system validates the credit card information
• The system issues a receipt to the customer

13
A. Use Case Point Method

• Classify each Actor and Use Case (as simple,
  average or complex). Provide reasons.
• Using tables obtain the total weighted sums of
  Actors and Use Cases Unadjusted Use Case
  Points, UUCP
• Of the 13 Technical Complexity Factors and 8
  Environmental Factors, identify the ones that are
  relevant to your design
• Using tables, compute the values of TCF and EF
• Obtain UCP by multiplying UUCP with TCF and EF
• To obtain the Effort in (Product-Hours) multiply
  UCP with the Productivity Factor per UCP.
• (This varies from 10 hours to 40 Hours per UCP.
  The variation is due to factors that have not
  been taken into account such as nature of the
  team, the project, and the processes used by the
  organization)

14
A. Use Case Point Calculations -1
back
Step 1. Tabulate Actors Obtain UAW, Unadjusted
Actor Weights Step 2. Tabulate Use Cases
Obtain UUCW, Unadjusted Use Case
Weights Step 3 Total Unadjusted UCP,
UUCP UAW UUCW
15
A. Use Case Point Calculations -2

• Step 4 Identify grade Complexity Factors
  Obtain TCF
• TCF 0.6 (0.01
  TFACTOR)

16
A. Use Case Point Calculations -3

• Step 5 Identify grade Environment Factors
  Obtain EF
• EF 1.4 (-0.03 EFACTOR)
• Step 6 UCP UUCP TCF EF
• Step 7 Effort in Person-Hours UCP
  Productivity Factor
• To obtain effort in Person-Months, uEffort
  Effort/200

17
B. Function Points Albrecht, IBM

• Function Points measures size based on the
  functionality. Estimation for Data Processing
  designs requires the counting of
• External inputs those items provided by the
  user that describe distinct application-oriented
  data (such as file names and menu selections)
• External outputs those items provided to the
  user that generate distinct application-oriented
  data (such as reports and messages, rather than
  the individual components of these)
• External inquiries interactive inputs
  requiring a response
• External files machine-readable interfaces to
  other systems
• Internal files logical master files in the
  system

18
Unadjusted FPs for DP Designs
back
Ref COSMOS http//www.cs/umt.edu/RTSL/SoftwEng/c
osmos41
19
FP Technical Complexity Factors
back

• Data Communication
• Distributed Data Processing
• Performance Criteria
• Heavily Utilized Hardware
• High Transaction Rates
• Online Data Entry
• Online Updating
• End-user Efficiency
• Complex Computations
• Reusability
• Ease of Installation
• Ease of Operation
• Portability
• Maintainability

20
FP Degree of Influence
back

• Compute the Degree of Influence, DI based on
  selection from the list of 14 Technical
  Complexity Factors, TCF.
• Identify the factors relevant for your project
  and assign a grade (0 to 5) to each of the
  factors selected.
• DI Sum of the ratings of the selected factors

21
FP Applying TCF correction
back
14 ? 1

• TCF0.650.01 (Degree of Influence)j
• MinTCF0.65, MaxTCF1.35
• Function Points FP UFP TCF

22
KLOCs corresponding to Function Points
back

• Estimate for the number of KLOCs (Kilo Lines of
  Code)
• Assuming LOCs per FP for
• Java 53
• C 64
• aKLOC FP LOC_per_FP / 1000

23
C. Object Oriented DesignsSize Calculations
back

• C1. Class Based Designs
• C2. Rational Rose Real Time Designs

24
C1. Class based Designs

• Identify and classify all Classes (Boundary,
  Entity)
• For each class, identify and count all methods
  needed to complete the design. Use Sequence
  Charts if necessary
• Compute LOC (C)
• - For Boundary Classes
• LOC 13.20 Num of Methods
• - For Entity Classes
• LOC 18.25 Num of Methods
• bKLOC Sum of LOCs for all Classes / 1000

25
C2. Rational Rose Real Time Designs

• A Rational Rose Real Time, (RRRT) design consists
  of a hierarchical arrangement of Capsules each
  associated with its specific State chart.
• Complexity arises due to the existence of the
  number of elements (States, Transitions, Ports
  and the Protocols) in the capsule.
• If we were to write the code ourselves (instead
  of letting the RRRT system generate code for us)
  then assuming one method (code segment) to be
  associated with each element, the total number of
  lines per Capsule can be
• LOC per Capsule
• 18.25 ? of (States, Transitions, Ports,
  Protocols)
• bKLOC Sum of LOCs for all Capsules / 1000
• (Note If you were to utilize the code generating
  capability of RRRT then you have to furnish the
  code for the transitions only).

26
C2.Size Tabulation for RRRT Designs
back
27
Step 3. Compute Effort LOC ? Effort in PM
back

• 1. Identify the type of your project
• Organic Small teams, no formal processes
• Semi-detached Average size org, some processes
  are defined
• Embedded Mature organization with formal
  processes
• 2. Computer the effort in Person-Months using
  appropriate COCOMO multipliers and coefficients
  from the table below
• Effort Person Months, PMA(KLOCB)

28
Step 3Effort Adjustment, EAF Table
back
29
Step 3 Applying EAF correction to Person Months
PM ? APM
back

• Total EAF P (selected factors)
• Adjusted Effort Adjusted Person Months
• APM (Total EAF) PM

30
Step 4. Computing Time for Development Adjusted
Effort ? Time
back

• Compute the estimated optimal development time
  according to COCOMO.
• Development Time Equation Parameter Table
• Development Time
• TDEV C(APMD)

31
Step 5. Computing Number of Personnel APM/TDEV ?
NP
back

• Calculate the Number of Personnel NP
• (based on the Adjusted Effort in Person-Months
  and the duration of the project).
• NP APM / TDEV

32
Step 6. Distribution of Effort Process Phases
back

• A development process typically consists of the
  following stages
• Requirements Analysis
• Design (High LevelDetailed)
• Implementation Coding
• Testing (UnitIntegration)
• For the Project to be completed on time, it is
  important to allocate and budget for appropriate
  effort and time at each stage. If we do not do
  this the project may be over-budget, late or both.

33
Step 6. Distribution of Effort Process Phases
back

• The following table gives the recommended
  percentage distribution of Effort (APM) and TDEV
  for these stages
• Percentage Distribution of Effort and Time Table

34
Step 7. Estimation of Errors
back

• Calculate the estimated number of errors in your
  design, i.e. total errors found in requirements,
  specifications, code, user manuals, and bad
  fixes
• Update the Function Points calculated in step1 to
  EFP
• EFP FP 1.25
• Use the following table for calculating error
  estimates

35
D. MiscellaneousMeasures of Size Effort

• Feature Points
• For applications involving extensive use of
  algorithms and light use of files.
• FP 3Algorithms 3Master Files
• Design Weights Demarco
• Powerful methodology based on the number of
  Decisions (If Case) statements and the
  input/output Tokens.

36
D. COCOMO II, Post-Architectural Model for Effort
back

• Effort Person Months, PMA(KLOCB)
• No notion of Development Modes (OO, SD, ER)
• Parameter A 2.55
• Parameter B 1.01 0.01 Sum (Scale Factors)
• B(min) 1.06, B(max) 1.25

37
Flowchart Estimation Methodologies
38
Design
Classes2 Function Points Capsulesx Function
Points
Unadjusted Function Point (UFP table)
DI?ratings of selected factors
14 TCF0.650.01?(Factor)j
1 MinTCF0.65 MaxTCF1.35
Modify FPUFPTCF
LOC18.25Num of Methods (modify for RRRT)
TCF
 bKLOC? (LOCs all Classes, Caps)/1000
AFPFP1.25
Error Table Y
Compute Errors AFPY
KLOCMaxaKLOC, bKLOC
uEffort UCP Method
cEffort Person Months, PMA(KLOCB)
Effort, PMMaxcEffort, uEffort
Result
Adjusted PM APM(total EAF)PM
EAFProduct of selected factor
Effort Factors1-15
Development Time TDEVC(APMD)
Number of personnel NPAPM/TDEV
39
Summary
back

• From the Size Metric we have estimated Effort
  its distribution, Duration, Persons and Errors
• How good is estimation - 400
• Accuracy of estimation converges to the actual
  value as time proceeds
• Accuracy of estimation depends upon
• - Quality of the High Level Design
• - The counting process for baseline metrics
• - How well the correction factors have been
  chosen and applied