Step 2: Make a Step 2: Make a Causal Model Causal Model Farrokh Alemi Ph.D. Farrokh Alemi Ph.D. This research was funded by Grant RO1 HL 084767 from the National Heart Blood and Lung Institute.
Jan 07, 2016
Step 2: Make a Causal Step 2: Make a Causal ModelModel
Step 2: Make a Causal Step 2: Make a Causal ModelModel
Farrokh Alemi Ph.D.Farrokh Alemi Ph.D.
This research was funded by Grant RO1 HL 084767 from the National Heart Blood and Lung
Institute.
Step 2: Make a Causal Step 2: Make a Causal ModelModel
Step 2: Make a Causal Step 2: Make a Causal ModelModel
This lecture continues from the This lecture continues from the lecture on making a list of lecture on making a list of
causescauses
Step 2: Make a Model from Your List
Put Each Cause in a Node
Put the Effect to the Right
Connect Causes to Effect
Put in Constraints
Equivalent Probability Model
Each Node a Function of Nodes Linking to It
Combination of Multiple Causes
p(Exercise | Ready to bike, Ready to shower at gym, Sleep early)= highp(Exercise | Ready to bike, Ready to shower at gym, Slept late)= Between low and high p(Exercise | Ready to bike, No plans to shower at gym, Sleep early)= Between low and high p(Exercise | No plans to bike, Ready to shower at gym, Sleep early)= Between low and high p(Exercise | Ready to bike, No plans to shower at gym, Slept late)= Between low and high p(Exercise | No plans to bike, Ready to shower at gym, Slept late)= Between low and high p(Exercise | No plans to bike, No plans to shower at gym, Slept late)= Low
Out of Sight, Out of Mind
Breakfast time
Probabilistic Dependence
Probabilistic Dependence
biking)Plan |Exercise(Exercise)( pp
Probabilistic Dependence
Probabilistic Dependence
shower) gymPlan |Exercise(Exercise)( pp
Probabilistic Dependence
early) Sleeping|Exercise(Exercise)( pp
Probabilistic Independence
early) Sleeping|Exercise(Exercise)( pp
Probabilistic Independence
early) Sleeping|shower Plan to(shower) Plan to( pp
Serial Nodes: Root Causes
Rain) bike, Ready to|Exercise(
bike) Ready to|Exercise(
p
p
Multiple Causes
early) Sleeping Exercise,|shower Plan to(
Exercise)|shower Plan to(
p
p
Probabilities Can Be Tested Against Data
Lecture ContinuesLecture ContinuesLecture ContinuesLecture Continues
Step 3: Thought ExperimentsStep 3: Thought Experiments