Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Module 5 Bayesian belief network modelling

1,690 views

Published on

Bayesian belief network modelling

Published in: Education
  • Login to see the comments

Module 5 Bayesian belief network modelling

  1. 1. BAYESIAN BELIEF NETWORK MODELING Professor Ockie Bosch Dr Nam Nguyen
  2. 2. A Reminder: Where does BBN modeling fit into the Evolutionary Learning Laboratory for Managing Complex Issues? New Levels of Learning and Performance at Local Level 1. IDENTIFY ISSUES 4. IDENTIFY LEVERAGE/ SYSTEMIC INTERVENTIONS 3. DEVELOP OR REFINE SYSTEMS MAPS OR MODELS 7. REFLECTION Environmental Economic Social 5. INTEGRATED SYSTEMIC MANAGEMENT PLAN 2. BUILD CAPACITY 6. IMPLEMENT STRATEGIES, POLICIES Stakeholders’ Mental Models Systems Structure Patterns & Relationships Cultural Values Pollution + Temporary immigration Empoloyment opportunity Land required for tourism Natural beauty - Availability of underground water + Fresh water consumption Employment opportunity for local people + + Social issues Infrastruture & facility + + Number of tourists Attractiveness of Cat Ba Island - + - Wealth of local people + + Waste + + + Illegal forest - exploitaion - Total population Agricultural Production Living cost - + + - + - + + - Conservation and agricultural land + - B2 R1 B1 R3 R2 B3 B4 B5 B6 Student population Assess to education + - - - R5 R4 Investment + + + Resident population + + BAYESIAN NETWORK MODELS © Professor Ockie Bosch and Dr Nam Nguyen
  3. 3. Why using Bayesian Belief Network Models?  Useful for Participatory Systems Analysis  Easy to construct for non-modellers - Visual  Ideal mechanism for integrating mental models, different types and forms of knowledge and data  Can link qualitative and quantitative factors in one model  They represent complex systems (i.e. link data, information and processes)  Quantify uncertainties through the use of probabilities  Easily updateable – through adaptive management/learning by doing  Determine what to do systemically (not just trial and error)  Can rapidly perform diagnostic and sensitivity analysis; Scenario testing  Useful for developing strategic and operational plans © Professor Ockie Bosch and Dr Nam Nguyen
  4. 4. What is a Bayesian Network? Nodes (Variables) Rain Temp Links (Representing causal relationships) Growth Representation of • Graph theory (structures) • Probability theory (strength of relationships between variables Inputs: • Expert opinion (technical / non-technical) • Literature findings • Monitoring & research data • Other models © Professor Ockie Bosch and Dr Nam Nguyen
  5. 5. IMPROVEMENT OF ….. Probability that systemic intervention will achieve the goal © Professor Ockie Bosch and Dr Nam Nguyen
  6. 6. Things You Can Do – Management Interventions A B IMPROVEMENT OF ….. Probability that systemic intervention will achieve the goal © Professor Ockie Bosch and Dr Nam Nguyen
  7. 7. B Effects or Outcomes of Management Interventions 1 2 3 A IMPROVEMENT OF ….. Probability that systemic intervention will achieve the goal © Professor Ockie Bosch and Dr Nam Nguyen
  8. 8. C D E F Other Controlling X Factors Y B 1 2 3 A IMPROVEMENT OF ….. Probability that systemic intervention will achieve the goal © Professor Ockie Bosch and Dr Nam Nguyen
  9. 9. INTEGRATED STRATEGIC MANAGEMENT PLAN Other Controlling Planning X Factors Y What needs to be in place? G H I J Systemic Interventions; Strategies; Activities; 1 2 3 Preparation A IMPROVEMENT OF ….. Probability that systemic intervention will achieve the goal © Professor Ockie Bosch and Dr Nam Nguyen B CC DD EE FF A B X Y Projects; Timing
  10. 10. Probability to improve SDM into a world class institution © Professor Ockie Bosch and Dr Nam Nguyen
  11. 11. Probability to improve SDM into a world class institution © Professor Ockie Bosch and Dr Nam Nguyen
  12. 12. Quality of Life of Ghana women in agriculture Willingness_to_learn Yes No Work_Pressure Reduced Increased 48.3 51.7 Qual_of_Life_WISSA High 37.9 Low 62.1 Knowledge_Skills Improved Unchanged 47.3 52.7 Family_Support Yes No 36.9 63.1 Infrastructure Modern Poor 45.8 54.3 Gov_Support Effective Ineffective 50.0 50.0 Produce_Quality High Poor 29.9 70.1 Market_actors_Linkages Strong 36.0 Weak 64.0 Producer_Groups Strong Weak 15.0 85.0 Branding Yes No 23.9 76.1 Market_Access Good Poor 36.8 63.2 Satisfied_Consumers Yes 33.4 No 66.6 Stable_Markets Yes No 37.4 62.6 Income High Low 32.7 67.3 Quality_control Effective Ineffective 15.0 85.0 Secondary_Jobs Available Unavailable 18.0 82.0 Produce_Prices High Low 28.7 71.3 Eco_friendly_Practices Yes 30.0 No 70.0 Toxic_Chemicals_Use Reduced 29.7 Increased 70.3 Rural_hygience Clean Poor 25.0 75.0 Health Good Not_good 26.9 73.1 Service_groups Effective Ineffective 66.5 33.5 Implements Available Unavailable 25.3 74.8 Prod_efficiency High Low 51.5 48.5 Input_prices_quality Satisfactory Unsatisfactory 30.8 69.2 Access_Healthcare Yes 30.9 No 69.1 Healthcare_Services Effective 35.0 Poor 65.0 Training_content_relevance Yes 40.0 No 60.0 Capital Available Insufficient 10.0 90.0 Qual_Training Yes No 39.8 60.2 Old_Customs Yes No 10.0 90.0 70.0 30.0 Social_Engagement High 13.0 Low 87.0 Learning_Capacity High 60.0 Poor 40.0 Qual_Trainers Good Poor 55.0 45.0 Production_Cost Reduced Expesive 35.8 64.2 Product_Diversification Yes 40.0 No 60.0 Work_sharing Yes No 38.7 61.3 © Professor Ockie Bosch and Dr Nam Nguyen - Current situation
  13. 13. Systemic interventions identified Willingness_to_learn Yes No Work_Pressure Reduced Increased 78.2 21.8 Qual_of_Life_WISSA High 75.8 Low 24.2 Knowledge_Skills Improved Unchanged 65.7 34.3 Family_Support Yes No 83.0 17.0 Infrastructure Modern Poor 90.0 10.0 Gov_Support Effective Ineffective 100 0 Produce_Quality High Poor 82.3 17.7 Market_actors_Linkages Strong 100 Weak 0 Producer_Groups Strong Weak 100 0 Branding Yes No 65.8 34.2 Market_Access Good Poor 76.8 23.2 Satisfied_Consumers Yes 64.5 No 35.5 Stable_Markets Yes No 95.0 5.00 Income High Low 78.5 21.5 Quality_control Effective Ineffective 15.0 85.0 Secondary_Jobs Available Unavailable 100 0 Produce_Prices High Low 73.4 26.6 Eco_friendly_Practices Yes 100 No 0 Toxic_Chemicals_Use Reduced 82.2 Increased 17.7 Rural_hygience Clean Poor 25.0 75.0 Health Good Not_good 63.1 36.9 Service_groups Effective Ineffective 73.0 27.0 Implements Available Unavailable 100 0 Prod_efficiency High Low 87.5 12.5 Input_prices_quality Satisfactory Unsatisfactory 95.0 5.00 Access_Healthcare Yes 67.1 No 32.9 Healthcare_Services Effective 100 Poor 0 Training_content_relevance Yes 40.0 No 60.0 Capital Available Insufficient 100 0 Qual_Training Yes No 56.0 44.0 Old_Customs Yes No 10.0 90.0 100 0 Social_Engagement High 100 Low 0 Learning_Capacity High 60.0 Poor 40.0 Qual_Trainers Good Poor 100 0 Production_Cost Reduced Expesive 83.9 16.1 Product_Diversification Yes 100 No 0 Work_sharing Yes No 78.4 21.6 © Professor Ockie Bosch and Dr Nam Nguyen
  14. 14. How to populate the model Demonstration on Follow steps on you computers
  15. 15. Live Demonstration with Netica & Hands-on I. What can be done to achieve the goal (address the goal)? (3 Max. per group) II. Why would you take these actions (what will be the outcomes that will lead to achieving the goal)? (2-3 Max. per action) III. What factors could influence the outcomes mentioned under point 2? IV. What need to be in place for the actions to happen? (2 – 3 Max. per action) © Professor Ockie Bosch and Dr Nam Nguyen

×