Biostatistical Computing, PHC 6068

Overview

Zhiguang Huo (Caleb)

Monday August 21, 2017

Jobs: Industry, academic, collaboration…
- Google’s Chief Economist: “I keep saying that the sexy job in the next 10 years will be statisticians. And I’m not kidding.”
Data analysis:
- Independence: You don’t rely on someone else giving you exactly the right tool.
- Honesty: You don’t distort your problem to match the tools you have.
- Clarity: You can turn your ideas into something a machine can do refines your thinking.
Fun: You can apply cool tools to convert data into knowledge.

Lecturer Zhiguang Huo (Caleb)
TAs: NA
Lecture times:
- Monday: 12:50-2:45pm, CTRB, Room 5235
- Wednesday: 1:55-2:45pm, CTRB, Room 5235
Office hours:
- Monday 3-5pm, CTRB, Room 5230
Course Website:

Major targeted audience:
- First year PhD students in Biostatistics department.
- Second year Master’s students in Biostatistics department.
No programming knowledge assumed
Some statistics knowledge assumed
- PHC 6092: Biostatistical Theory
- PHC 6050: Statistical Methods for Health Sciences Research I
- PHC 6051: Biostatistical Methods II
Focus almost entirely on R, will be some algorithm implementation.
Class will be cumulative, so keep up with the material and assignments!

No textbook is required
Recommanded reading:
- An Introduction to R. (From R official website). https://cran.r-project.org/doc/manuals/R-intro.html
- Google’s R Style Guide. https://google.github.io/styleguide/Rguide.xml
- Hadley Wickham (2014). Advanced R. Taylor & Francis. http://adv-r.had.co.nz

Format: will be released/submitted in R markdown.
Due time: Two weeks after the homework is released, Due on Mon 11:59 pm.
Late homework: 80% credit within one day late; 50% credit within two day late; 0% after two day’s late.
Where to submit: electronically via elearning.
Can I work with other people? Yes, You may work alone or with one teammate of your choosing. (Teams of three or more are not allowed.)

Mid-term exam: about 1 week take-home exam.
Final exam: about 5 days take-home exam.
Format: will be released/submitted in R markdown.
Due time: See course website, no late homework.
Can I work with other people? Yes, You may work alone or with one teammate of your choosing. (Teams of three or more are not allowed.). But you are not allowed to discuss with any other people except for your teammate.

Wrap up a R package (with one of these options)
Complete documentation of all your functions.
Apply R package to real data or synthetic data.
Report using R markdown, include how you used your own package.
How to make an outstanding R package?
- S3/S4 class
- calling c++
- make nice vignette
Instructions: will be released on Mon Oct 16.

All 3 are free, and all 3 will be used extensively in this course

(survey, demo)

All R materials are download from the ``Comprehensive R Archive Network" (CRAN), which is a collection of sites from around the world carrying identical and current R code and documentation.
R can be installed on Windows, Mac or Linux operating systems.
In this course, the instructors will run R on Mac.
To install R, visit the R website, http://www.r-project.org and follow the installation directions (see the FAQ section for more details).

When you first open R, you will see the R window. The console is the fundamental way for interacting with R.
In the console is a prompt, > this is where an expression, phrase of code that can be executed, is inputed into R (if entering lengthy expressions that require more than one line the prompt changes from > to +.
An expression is typed into the console at the prompt, press Enter, and the expression is evaluated.
Depending on the expression, the system may respond by outputting results to the console, creating a graph in a new window or performing some other task. Then another expression can be inputed and evaluated.
An R session is this interaction between you and the system.
Often included with outputted results in the console are numbers in [ ]. These numbers are used to reference locations in the output.

RStudio is a free and open source integrated development environment (IDE) designed specifically for R. http://www.rstudio.com
Runs on, Windows, Mac or Linux operating systems.
Useful productivity tools：
- Syntax highlighting, smart indention, parentheses matching
- Easily execute code from the source editor
- Integrates all of the tools (source editor, console, plots, documentation, and more) when working with R into a single environment
- Extensive tools for developing R packages
- Interactive debugging for diagnosing and fixing errors

Will be covered in the next lecture.

It’s free, open source, and available on every major platform.
A massive set of packges for statistical modeling, machine learning, visualization and importing and manipulating data.
Cutting edge tools.
A fantastic community.
Powerful tools for communicating your results. (html, pdf, interactive website).
An IDE (RStudio).
Connect to high-performance programming languages (C, C++, Java)
When you published a statistical methodology paper, often you will be asked to publish a R package as well.

R is not a particularly fast programming language, and poorly written R code can be terribly slow. R is also a profligate user of memory.
R language focuses on results instead of processes. Too many R functions use tricks to reduce the amount of typing at the cost of making code hard to understand.
R is slow in loop (e.g. for, while) so try to avoid loop in R if you run a large program and care about speed.