https://www.coursera.org/learn/model-thinking
Week 1
- Models tie us to a mast of logic
- Models should be blended with experience to make decisions.
- Models won't always work in all situations. Using lots of models can help inform when models are inappropriate
- Step 1. Name the parts (likely relevant parts only)
- Step 2. Identify relationships between parts
- Models can be used to inductively explore
- Models can be used to understand class of outcome (Cyclical, random, equilibrium, complex)
- Models can be used to understand patterns in data or predict points, produce bounds, and inform data collection
- Models can be used to identify and rank levers (imagine carbon cycle, ag is largest contributor, so it is a good lever)
- Models can be (1) equation based, (2) Agent-based - models interactions of agents based on behavior rules
-
Week 2
- Simple, rule based models can develop into equilibrium, patterns, chaos, or complexity
- Six sigma is a business process that is intended to ensure that failure only happens in 'six sigma' events, or where an event with a 3.4 in a million chance occurs.
-
Week 5
Markov Models
- Understanding conservation using Markov models
- The State and Transition Model is a Markov model for understanding ecological condition.
- Markov models are models that find equilibrium based on transition probablities
- Markov models will always arrive at the same equilibrium state given fixed transition probabilities
- A key lesson from Markov models is that interventions that only change some states to other states (e.g., restoration) do not affect the outcome--the equilibrium will always be achieved given enough time
- Interventions that only change some states to other states are sometimes useful because reaching the equilibrium may take time, and there may be some benefit achieved in the interim
- Truly impactful interventions are those that change the transition probabilities
- Thus, one-off restoration projects and similar efforts will only delay the inevitable
- True conservation solutions must seek to change the transition probabilities
- For example, sage-grouse habitat can be categorized into a number of states (A, B, C, D, E)
- Given the status quo, the transition probabilities are fixed and there is a shuffling of lands between states (e.g., some percent chance of overgrazing, some percent chance of development, some percent chance of restoration).
- With energy development, the transition probability from suitable to unsuitable increased
- To offset this impact, a similar change in transition probability from unsuitable to suitable must be achieved
- The offset must be sufficient to maintain the same equilibrium point