Please remove this section when submitting your homework.
Students are encouraged to work together on homework and/or utilize advanced AI tools. However, sharing, copying, or providing any part of a homework solution or code to others is an infraction of the University’s rules on Academic Integrity. Any violation will be punished as severely as possible. Final submissions must be uploaded to Gradescope. No email or hard copy will be accepted. For late submission policy and grading rubrics, please refer to the course website.
HWx_yourNetID.pdf. For example,
HW01_rqzhu.pdf. Please note that this must be a
.pdf file. .html format
cannot be accepted. Make all of your R
code chunks visible for grading..Rmd file
as a template, be sure to remove this instruction
section.R is \(\geq
4.0.0\). This will ensure your random seed generation is the same
as everyone else. Please note that updating the R version
may require you to re-install all of your packages.The first three questions of this homework is quite simple. We will
use three different models to estimate the personalized treatment rule,
and we will use the Sepsis dataset in all three questions.
Keep in mind that the Sepsis dataset contains a
BEST variable which is the doctor’s best guess for the
treatment. And you should never use this variable in your model fitting.
Only use it when evaluating the performance of your model. The data is
provided on our course website.
Implement the direct learning approach with linear regression. Since
this is an observational study, you should incorporate propensity score
weighting when estimating the treatment rule. After obtaining the
estimated treatment rule, compare it with the best label
Sepsis$BEST and report your results.
Use the outcome weighted learning via WeightSVM package
with support vector machine to analyze the Sepsis dataset.
You may have to consider the following decisions try some of them
If the performance is not as good as you want, try the
DTRlearn2 package to perform this task again. Note that the
DTRlearn2 package has already incorporated some rough
internal tuning procedures and you only need to try different kernels.
Report your results.
grf packageUse the grf package to find the optimal treatment rule
for the Sepsis dataset. Compare the estimated treatment
rule with the best label Sepsis$BEST and report your
results. Furthermore, report the estimated average treatment effect
(ATT) of this entire population (based on our understandings of the
relationship between the conditional average treatment effect and the
average treatment effect). You may also use the grf package
documentation.
Suppose that I want to estimate \(E(X^3)\) when \(X\) is a random variable from \(N(1,1)\). However, I do not know how to perform integration, and I could do is to generate random samples and perform numerical approximation. In addition, my machine can only generate random samples from \(N(2,2)\), but I do know the analytic density functions of both distributions. How can I estimate/approximate \(E(X^3)\) numerically using the Radon–Nikodym theorem? Use 5000 samples to make your estimation stable. Validate your answer using simulated data from \(N(1,1)\) or analytic integration.
Hint: if you ask GPT or write your prompt correctly with Copilot, it will give you the answer. But make sure to explain your procedure.