You are on page 1of 10

Case Study: Version 2.

0 – adoption rules
• Add broadcast-influence and social-influence sliders
to the model
• Have agents adopt based on the Bass model so agents
adopt either due to:
• broadcast-influence
• social-influence * fraction-of-adopters
• Optional: Have different adoption rules?

Master of Business Analytics 7 November 2020 28


Case Study: Version 2.0 – adoption (contd.)
• According to multiple studies broadcast influence is
quite low compared to social influence
• Both are probability values between 0 and 1 (step?)
• So broadcast-influence slider can default to 0.01
• While the social-influence slider can default to 0.5
• But actual social influence depends on current adoption:
social-influence * fraction-of-adopters
• Write two separate rules for these influences in the code –
where? How to compute this?

count turtles with [adopted?]


count turtles
Master of Business Analytics 7 November 2020 29
Case Study: Version 2.0 – adoption (contd.)
• Complete the code

• What happens when run?


• How to avoid keeping re-adopting?

Master of Business Analytics 7 November 2020 30


Case Study: Version 3.0 – output monitoring
• Just visualizing the population is insufficient
• Need a better way to visualize the result
• Plot the Adoption Rate over time
• To see if adoption curves are what we expect (which is?)
• Stop the model if no one is left to adopt
• Optional: Plot the different types of adoption
separately
• Due to broadcast influence
• Due to social influence

Master of Business Analytics 7 November 2020 31


Case Study: Version 3.0 – output (contd.)
• Add the Plot widget in the UI
• Name the plot, the x-axis and y-axis
• We need to plot the number of turtles who have adopted
• Stop the model if no one is left to adopt
• Need to check at the top of the go procedure
• Something like: any? turtles with [adopted?]
• Things to do:
• Slow down if needed
• See how many ticks before stop
• What is the shape of the curve?
Master of Business Analytics 7 November 2020 32
Case Study: Version 4.0 – social connections
• Add a Random Network to the model
• To simulate what happens in the real world!
• Adoption Rule Change
• Rather than social-influence being modified by thewhole
population of adopters, it is only modified by the local
neighborhood
• Optional: Set up a distribution for the density of
agents

Master of Business Analytics 7 November 2020 33


Case Study: Version 4.0 – connections (contd.)
• In ABMs we try to avoid global information
• But so far, all agents are influenced the same way!
• We’re influenced by our ‘social neighbors’ (who are they?)
• Can ask agents to create-links-with others
• Undirected links (work both ways)
• create-links-from and create-links-to are
unidirectional
• Can use show count links in command center to see
how many links were created – check it

Master of Business Analytics 7 November 2020 34


Case Study: Version 4.0 – connections (contd.)
• This doesn’t change the way agents get influenced
• Need to modify social influence procedure
• Broadcast influence is not affected in general (no?)
• The numerator and denominator in the social
influence calculation have to change
• It should be the number of neighbors who have adopted
over the total number of neighbors for each agent
• Use link-neighbors to access the number of neighbors
• How many neighbors can each agent have?

Master of Business Analytics 7 November 2020 35


Case Study: Version 5.0 – reruns
• How to measure effects of parameters for a single
network configuration?
• Current rerun, destroys the world!
• Add a reset innovation button so you can run on the
same network multiple times
• Call it reset-diffusion?

Master of Business Analytics 7 November 2020 36


Case Study: Version 5.0 – reruns (contd.)
• What are the things we need to reset?
• All turtles have adoption set to true – change to false
• All turtles need to be repainted white
• Try it out!
• Can try different values for social influence and check
patterns
• How could you reset the graph?
• Hint: use clear-all-plots (where?)
• If you only wanted to clear one plot?

Master of Business Analytics 7 November 2020 37

You might also like