ST 495 Final project outline:


Building on the data analysis and simulation study from your midterm project,
reconstruct your simulation study from a bootstrapping approach.

Determine an appropriate strategy for bootstrapping your real data set based on 
the model(s) you considered for your midterm project. 

Investigate the bootstrapped sampling distributions of your estimators, and 
comment on how inference from these sampling distributions compares to the 
parametric inference you presented in you midterm project.  For example, are the
bootstrapped confidence intervals similar to your parametric confidence intervals? 



General outline of the bootstrap study:

(1) Generate 1 bootstrapped data set from the real data and statistical model
(2) Implement an estimation procedure (i.e., MLE, EM, MCMC,...)
(3) Repeat steps (1)-(2) N times (larger N is better, but may not be feasible)
(4) Use the output from (3) to produce plots and/or tables as in the paper

Organize the simulation study into 3 script files:
(a) bootstrap.r  # This script contains code for bootstrapping the data 
(b) out_file.r  # This script contains code for producing the plots/tables
 
The output from bootstrap.r should be saved and then loaded by out_file.r. And,
the output from out_file.r should be saved, and then loaded (or copy-and-pasted)
into your LaTex file to be compiled into the pdf file of the write-up of your 
project.

Next, your project submission should include a workflow file, which includes 
ALL of the steps for reproducing your computations and plots/tables.  For 
example, it will look something like:

-------------------------------------------------------------------------------
# This is the workflow to reproduce the results of the ST 495 midterm project by
# (student name)

for seed in {1..N}
do
Rscript bootstrap.r arg_1 arg_2 ...
done

Rscript out_file.r arg_1 arg_2 ...
-------------------------------------------------------------------------------

Note: This will require knowledge of how an interpretable language works 
outside of a GUI environment (i.e., you should know how to run R outside of 
Rstudio).  This is not hard, but is important for understanding how programming 
skills fit into the broader application of problem solving in industry and 
academia.  

Finally, your project write-up will include a 10 page summary describing the
statistical model, estimation procedure, and inference on your bootstrap study.
You will also need to send me all script files INCLUDING your workflow file.