Docstoc

Project Management Theory of Probability

Document Sample
Project Management Theory of Probability Powered By Docstoc
					                                                                                                                                                                                                                                2
                                                                                                                      Agenda - What will you learn today?                                                        Kristian.Sandahl@
                                                                                                                                                                                                                              liu.se



                                                                                                                                       Part I                                                 Part III
                                                                                                                         Follow-up from last lecture and                 More on Prioritization and Estimation

                    Project Management                                                                                                exercise



                                           SEMINAR

                                    Software Engineering
                                     TDDC88/TDDC93
                                          autumn 2010


                                Krisitan Sandahl, David Broman
                          Department of Computer and Information Science
                                  Linköping University, Sweden
                                     Kristian.Sandahl@liu.se

                                                                                                                         Part I                       Part II
                                                                                                                         Follow-up                    Estimation and
                                                                                                                                                              prioritization




                                                                                                              3                                                                                                                 4
                                                                                                Kristian.Sandahl@     Last lecture                                                                               Kristian.Sandahl@
                                                                                                             liu.se                                                                                                           liu.se




                                                                                                                           Definition of a project                           Buffer time
                                                                                                                                                                       
                              Part I                                                                                    
                                                                                                                            Stakeholders
                                                                                                                            SMART goals                                 
                                                                                                                                                                               Resources
                                                                                                                                                                               Roles
                                                                                                                           Project parameters                                Risk management:
                       Follow-up from last                                                                                 GANTT chart                                        
                                                                                                                                                                               
                                                                                                                                                                                   Identification
                                                                                                                                                                                   Analysis
                                                                                                                               Task (activity)
                       lecture and exercise                                                                                   
                                                                                                                              
                                                                                                                                Duration
                                                                                                                                Precursor (predecessor)
                                                                                                                                                                               
                                                                                                                                                                               
                                                                                                                                                                                   Planning
                                                                                                                                                                                   Monitoring
                                                                                                                               Phase                                    Probablility x Impact = Risk
                                                                                                                               Critical path                             magnitude indicatior
                                                                                                                               Available time = Slacktime +             Kick-off
                                                                                                                                Real time
                                                                                                                               Milestone
                                                                                                                               Tollgate




    Part I                           Part II                                                                             Part I                       Part II
    Follow-up                        Estimation and                                                                      Follow-up                    Estimation and
                                             prioritization                                                                                                   prioritization




                                                                                                              5                                                                                                                 6
3. Risk Planning                                                                                Kristian.Sandahl@     The Project Plan - Content                                                                 Kristian.Sandahl@
                                                                                                             liu.se                                                                                                           liu.se

                                                 "Communication problem between develop sites in
                                                       Stockholm and India                                            Project Description                                      Time and Resource Plan
     1. Risk Avoidance                            - localize all development in India?"                                  Background to the project                               Milestones
                                                                                                                         Relevant constraints (budget etc.)                      Tollgates
                                                   "the web-server fails often - low accessibility                       Project Goal                                            Deliverables
Reorganize so that the risk disappears.             outsource the operation?"                                            Start and expected end date.                            Activities
                                                                                                                                                                                  Resources
       2. Risk Transfer                              "Changes of requirements late in project
                                                      - a prototype?"                                                 Project Organization                                     Training Plan
Reorganize so that someone else takes                                                                                    Roles                                                    Needed knowledge and skills.
    the risk, insurance, customer, bank.                                                                                 Knowledge / skill                                        Who needs what? Budget?
                                                        Mitigate the risk        Lower the
                                                                                     probability.                        Communication and reports
    3. Risk Acceptance                                                                                                                                                         Change and configuration
                                                      "The key architect starts to work for another                   Risk Management                                                management
                                                                            company - 2 architects?"                     Risks, Probability and Impact                        (In larger projects, this part is a
     Live with it                                       Define                                                                                                                       document of its own.)
                                                                                A plan B...
                                                                                                                         Mitigation and Contingency plan
                                                   Contingency plan

    Part I                           Part II                                                                             Part I                       Part II
    Follow-up                        Estimation and                                                                      Follow-up                    Estimation and
                                             prioritization                                                                                                   prioritization




                                                                                                                                                                                                                                       1
                                                                                                        7                                                                                                                    8
Summary of project background                                                             Kristian.Sandahl@                                                                                                   Kristian.Sandahl@
                                                                                                       liu.se                                                                                                              liu.se




    3 destinations:                                    A new law proposal, my add req.
     Stockholm, London, Singapore                       Up and running with users within 1
    Now customer supp., manual                          year
     booking in London & Singapore                      Functional req. collected by asking                                      Part III
    Want: new booking sys with on-line                  users.

 
     booking + phone call booking
     3 dev – Microsoft .NET
                                                  
                                                  
                                                         Now, manual booking over phone
                                                         Hosting 3rd party, India
                                                                                                                    More on Priotization and Estimation
    1 Architect. 2 Modules: GUI,                       Customer database 25000
     Backend                                             customers
    Can complement with 2 roles                        Value for us: 4 man years.
    Integrated with an existing system
    Users are not experienced with
     computers
    Should be developed in PHP with
     MySQL as DBMS.



  Part I                        Part II                                                                               Part I                              Part II
  Follow-up                     Estimation and                                                                        Follow-up                           Estimation and
                                        prioritization                                                                                                            prioritization




                                                                                                        9
Prioritization of requirements                                                            Kristian.Sandahl@      Relative importance – Analytical Hierarchy Process                                           Kristian.Sandahl@
                                                                                                       liu.se                                                                                                              liu.se

                                                                                                                1. Expert pairwise comparison                                              3. Calculate normalized
                      Customer                                                                                                                       7                                     eigenvector = relative
                      Value                                                                                       F1: On line group-booking                      F3: Last minute tickets   importance

                                                                                                                                                             5
                                                                                                                                       3

                    High      Sweet Spot
                                                                                                                                           F2: Round-trip tickets
                                                                                                                                                                                             Approximation:

                                                                                                                2. Comparison matrix
                                                                                                                                             F1          F2                 F3
                                                                                                                                                                                                 F1             0.083
                                                                                                                                  F1         1           1/3                1/7
                     Low                                   Avoid                                                                                                                                 F2      =      0.193
                                                                                                                                  F2         3           1                  1/5
                                                                                                                                                                                                 F3             0.724
                                                                                                                                  F3         7           5                  1
                                Low                       High
                                                           Development Effort
                                                                                                                For enthusiasts: http://www.boku.ac.at/mi/ahp/ahptutorial.pdf
  Part I                        Part II                                                                               Part I                              Part II
  Follow-up                     Estimation and                                                                        Follow-up                           Estimation and
                                        prioritization                                                                                                            prioritization




                                                                                          Kristian.Sandahl@                                                                                                   Kristian.Sandahl@
                                                                                                       liu.se                                                                                                              liu.se




  Part I                        Part II                                                                               Part I                              Part II
  Follow-up                     Estimation and                                                                        Follow-up                           Estimation and
                                        prioritization                                                                                                            prioritization




                                                                                                                                                                                                                                    2
                                                                                               Kristian.Sandahl@     AHP usage                                                                 Kristian.Sandahl@
                                                                                                            liu.se                                                                                          liu.se




                                                                                                                      Pros:                                        Cons:
                                                                                                                       Multiple criteria                           Grows quadratically
                                                                                                                       Simple comparison                           Relative values only
                                                                                                                       Fast                                        Though chains of
                                                                                                                       Gives values on rational                     inconsistency
                                                                                                                         scale                                      Hard to add new alternatives
                                                                                                                                                                    Needs a tool




  Part I                          Part II                                                                              Part I                    Part II
  Follow-up                       Estimation and                                                                       Follow-up                 Estimation and
                                          prioritization                                                                                                 prioritization




Expert judgment – Planning Poker                                                               Kristian.Sandahl@     Planning poker                                                            Kristian.Sandahl@
                                                                                                            liu.se                                                                                          liu.se




  Each developer has a set of cards, usually with                                                                    Pros:                                        Cons:
   values: 0, ½, 1, 2, 3, 5, 8, 13, 20, 40, 100, ”don’t                                                                Each participant make own                   Need of tools for distributed
   know”                                                                                                                 decision                                    projects
  Values translate into days or “points”                                                                              All participants are active                 The orderer of the project
                                                                                                                       Participants get deeper                      needs to take consequences
  Requirement and user story is described
                                                                                                                         knowledge                                  Risk of over-rating
  All developers picks a card                                                                                                                                       confidence when too little
                                                                                                                       Iterative work
  All disclose their cards                                                                                                                                          information is present
                                                                                                                       Consensus based final
  Discussion, time-boxed, lowest and highest estimator                                                                  estimate
   start.                                                                                                              Fun
  New estimation round

Enthusiasts: http://www.planningpoker.com/

  Part I                          Part II                                                                              Part I                    Part II
  Follow-up                       Estimation and                                                                       Follow-up                 Estimation and
                                          prioritization                                                                                                 prioritization




                                                                                                            17                                                                                              18
Effort Estimation                                                                              Kristian.Sandahl@     Illustrating example, COCOMO                                              Kristian.Sandahl@
                                                                                                            liu.se                                                                                          liu.se




 Expert Judgment - the Delphi technique
                                                                                [No change]                            Effort = C1 EAF (Size)P1
      Experts make individual          Calculate                Mean is presented                                            Effort = number of staff months
        predictions secretly            Mean                     to expert group
                                                                                                                             C1 = scaling constant
                                                                                                                             EAF = Effort Adjustment Factor
                                  [An expert changes its estimate]
                                                                                                                             Size = number of delivered, human produced source code
                                                                                                                              instructions (KDSI)
 Algorithmic Methods - COCOMO and COCOMO II                                                                                  P1 = exponent describing the scaling inherent of the process
                                                 COCOMO II                                                                    (0.91-1.23)
  COCOMO (Boehm, 1981)
        An formula where parameters are                  Takes into account changes in SE,
         estimated using real projects.                    such as component reuse, prototyping
        Input: No of code lines                          Other inputs than number of code
                                                           lines. E.g. function-points (e.g.
        Output: Effort (time)
                                                           external in/out, user interactions, files)


  Part I                          Part II                                                                              Part I                    Part II
  Follow-up                       Estimation and                                                                       Follow-up                 Estimation and
                                          prioritization                                                                                                 prioritization




                                                                                                                                                                                                                     3
                                                                                     19                                                                             20
Illustrating example, COCOMO II                                         Kristian.Sandahl@     Algorithmic or parametric methods                        Kristian.Sandahl@
                                                                                     liu.se                                                                         liu.se




 Predict maintenance size:                                                                     Pros:                                Cons:
  Size = ASLOC *0.01*                                                                          Based on empirical                 Data collection
             Assessment and Assimilation (0-8) (effort to test other S/W)
                                                                                                 data                                planned and
             Software Understanding (10-50) (low:good structure)
             0.4 * percentage of changed design                                                Potential up to +/-                 perfomed
             0.3 * percentage of changed code                                                   20 % accuracy                      Expensive
             0.3 * percentage of integrated external code                                                                           consultants
                                                                                                No human bias
                                                                                                                                    Rapid change in
                                                                                                                                     technology


  Part I                         Part II                                                        Part I           Part II
  Follow-up                      Estimation and                                                 Follow-up        Estimation and
                                         prioritization                                                                  prioritization




                                                                                                                                                                             4

				
DOCUMENT INFO
Description: Project Management Theory of Probability document sample