Causal Inference Models and Data-Driven Methods
Examples
•
Causality interpretation
•
Unobserved heterogeneity
•
Endogeneity: Self Selectivity
Example:
Causality
interpretation
People
are
told
to
run
around
in
a
dark
room
for
5
minutes
Observation:
◦
Men
are
found
to have
many
more
head
injuries
than
women
Conclusion:
◦
Women
see
better
in
the
dark?
◦
Men
are
more
reckless
runners?
Example:
Unobserved
heterogeneity
Women
who
smoke
have
babies
that
are
600
grams
under
weight
on
average
Problem:
◦
Is
it smoking or unobserved
factors
that
are
correlated
with
smoking?
Example:
Endogeneity:
Self Selectivity
Effectiveness
of
Side-
Impact
Airbags
Cars
with
side
impact
airbags
have
lower
injury severities
Problem:
◦
People owning
side-impact
airbag are
not
a
random
sample of
the
population
(likely
safer drivers)
◦
Safer drivers would naturally
have lower injury severities
Example:
Endogeneity:
Self Selectivity
Effectiveness
of
Motorcycle
Safety
Course
People
who
take
motorcycle
safety
courses
have
higher
crash
rates
Are courses ineffective?
Problem:
◦
People
taking the course
are not a
random
sample
of
the
population
(likely
less
skilled)
Data
analysis
methods
Data driven methods:
•
Support Vector Machine
•
Random Forest
•
eXtreme Gradient Boosting (XGBoost)
•
Deep Neural Network
•
Recurrent Neural Network
•
Convolutional Neural Network
•
Agent-Based Machine Learning
•
Etc.
Data size and
prediction/causality
trade-offs
Prediction/causality/data size trade-offs:
Delve into various examples of causal inference models and data analysis methods, from traditional statistical models to cutting-edge data-driven approaches like AI/ML. Understand the challenges of causality interpretation and explore the trade-offs between data size, prediction, and causality in different scenarios.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
People around 5 5 minutes People are around in minutes are told in a a dark told to dark room to run room for run for Observation: Men are found to have many more head injuries than women Conclusion: Women see better in the dark? Men are more reckless runners? Example: Causality interpretation
Women babies under Women who babies that under weight who smoke that are weight on smoke have are 600 on average have grams average 600 grams Problem: Is it smoking or unobserved factors that are correlated with smoking? Example: Unobserved heterogeneity
Effectiveness Airbags Cars with side impact airbags have lower injury severities Effectiveness of Airbags of Side Side- -Impact Impact Problem: People owning side-impact airbag are not a random sample of the population (likely safer drivers) Safer drivers would naturally have lower injury severities Example: Endogeneity: Self Selectivity
Effectiveness Safety Effectiveness of Safety Course of Motorcycle Motorcycle Course People who take motorcycle safety courses have higher crash rates Are courses ineffective? Example: Problem: People taking the course are not a random sample of the population (likely less skilled) Endogeneity: Self Selectivity
Causal-inference models Data analysis methods Traditional statistical models Heterogeneity models Data-driven methods (AI/ML)
Data driven methods: Support Vector Machine Random Forest eXtreme Gradient Boosting (XGBoost) Deep Neural Network Recurrent Neural Network Convolutional Neural Network Agent-Based Machine Learning Etc.
Data size and prediction/causality trade-offs
Prediction/causality/data size trade-offs: Predictive Capability Causality/Inference Capability Big-Data Suitability Data-Driven Methods Heterogeneity Models Traditional Statistical Models Causal-inference models
