AMS 597: Statistical Computing
Pei-Fen Kuan (c)
Applied Math and Stats, Stony Brook University
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 1 / 9
Project Description
Your project involves creating an R package that performs the
following tasks.
You will work in pairs (groups of 2)
You can also opt out and work on the project individually
(Deadline to opt out is Thursday March 10 at 5PM EST by
emailing the instructor)
The instructor will make initial group assignment randomly this
Friday and post the pairing on Blackboard. You are allowed to
swap groups
Deadline for group swapping is March 24, 2022 at 10AM EST.
Email the instructor your new group, cc’ing the members of your
old and new group.
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 2 / 9
Project Description
Your R package will take as input a response variable y and matrix
of candidate predictors/independent variables X, where each column
is a predictor.
Your package will work for both binary y and continuous y (for
continuous case, it can be assumed to be normally distributed).
The number of predictors p can be large (i.e., you should also
consider the case where p > n, n is the sample size).
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 3 / 9
Project Description
Your package will allow user specifiy the type of model to fit:
I linear or logistic regression
I ridge regression (for binary and continuous y)
I lasso regression (for binary and continuous y)
I random lasso (for binary and continuous y)
For random lasso, you will read the following paper: “Random
Lasso”, Annals of Applied Statistics (2011)
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3445423/
You will implement the algorithm described in Section 2 (Algorithm
(“Generate” and “Select”)).
For Step 2b, you only need to implement for lasso (you do not need
to implement adaptive lasso).
The pdf version of this paper is available at Blackboard (see
ScientificPaper_ProjectSpring2022.pdf)
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 4 / 9
Project Description
Your function will fit a prediction model to the input data
You can either treat the entire data as training data, or you can also
make it more user friendly by allowing user to divide the data into
training/test, and evaluate the model performance on the test data.
Your package can import glmnet and use the functions in this
package.
You will then wrap these up as an R package called
extendedglmnet
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 5 / 9
Project Description
The R package has to be complete and contains a vignette
describing how to use the R package
The R package is due May 05, 2022 at 5:00 PM
Submit your package as original source package (i.e., .tar.gz file) on
Blackboard>Assignments>Project. Name your package
extendedglmnetGroupX_version.tar.gz, where GroupX is your
group number, e.g., Group1
Version is generated automatically after you build your package
successfully
All students will submit the R package to Blackboard (i.e.,
although both members will submit the same R package, I still
require both to submit to their respective Blackboard workspace).
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 6 / 9
Project Description
Some of the grading criteria include:
I Can the R package be installed successfully?
I Is the R package implementing the required method correctly?
I Has it considered all possible scenarios?
I Is the R package user friendly (vignette, help files, warning messages,
sample data, sample code)?
I What is the computational speed?
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 7 / 9
Project Description
Some useful links:
https://tinyheero.github.io/jekyll/update/2015/07/26/
making-your-first-R-package.html
https://hilaryparker.com/2014/04/29/
writing-an-r-package-from-scratch/
https://combine-australia.github.io/r-pkg-dev/
http://kbroman.org/pkg_primer/
http://kbroman.org/Tools4RR/assets/lectures/08_rpack_
withnotes.pdf
https://cran.r-project.org/doc/contrib/
Leisch-CreatingPackages.pdf
https:
//ourcodingclub.github.io/tutorials/writing-r-package/
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 8 / 9
Project Description
Some useful links for incorporating existing R package into your R
package
https://kbroman.org/pkg_primer/pages/depends.html
https://r-pkgs.org/description.html
Or google keywords import R package'',depends R package”
Some useful links (for Windows):
https:
//www.biostat.wisc.edu/~kbroman/Rintro/Rwinpack.html
Pei-Fen Kuan (c) (Applied Math and Stats, Stony Brook University)AMS 597: Statistical Computing 9 / 9