R_et_co.Rmd

---
title: "R and related tools for research and teaching"
author: "Jouni Helske & Satu Helske"
date: "4 December 2015"
output: ioslides_presentation
---
<!-- -->
## Working more efficiently with R

![kuva1][kuva1]

[kuva1]: kuva1.png

<!-- These cover a wide range of academic work from data manipulation and analysis to software development, reporting and teaching. We glance at topics such as reproducibility, version control, packaging, and testing. -->

## R

<!-- - More and more popular (also outside academia) -->
- More than 2 million users ([Oracle estimate](http://www.oracle.com/us/corporate/press/1515738) in 2012)
- 6th most popular programming language by [IEEE Spectrum](http://spectrum.ieee.org/computing/software/the-2015-top-ten-programming-languages) (in 2015)
- 3rd in number of scholarly articles in Google Scholar (in 2014; [blog post](http://r4stats.com/articles/popularity/) by R.A.Muenchen)
- UseR! 2015 conference: 660 participants, 284 from industry
- R activity around the world: http://rapporter.net/custom/R-activity
- https://www.r-project.org/
<!-- Gergely Daróczi; Switzerland 1st, Finland 18th -->

## Rstudio

- Open source graphical interface for R and related tools
- Code completion, syntax highlighting, code diagnostics
- Autosave, command history, plotting history, environment browser, integrated searchable help
- R projects for easy return to and switching between jobs
- https://www.rstudio.com/

## R Markdown

- Simple formatting syntax for authoring HTML and PDF documents and slides
- Based on Markdown language, `knitr`, and pandoc
- Fully reproducible documents
- Closely integrated to Rstudio (New File -> R Markdown)
- http://rmarkdown.rstudio.com/
<!-- This is an R Markdown presentation. -->

![Rmarkdown][Rmarkdown]

[Rmarkdown]: Rmarkdown.png 

## Rmd features

- Equations like in LaTex: 
  \$y_i = \\beta x_i\$ produces $y_i = \beta x_i$       
- Embedding R code in the document:
```{r}
summary(cars)
```

## Rmd figures
```{r}
plot(cars)
```

## Rmd tables
```
Type     | Freq  | %
---------| ------|-----
Apples   | 7     | 44
Oranges  | 9     | 56

```

Type     | Freq  | %
---------| ------|-----
Apples   | 7     | 44
Oranges  | 9     | 56


## R package creation

- Sharing work with others 
    + Co-workers, CRAN etc.
- Personal projects
    + Loading functions, data and other packages at once
- Easier integration with C/C++/Fortran codes with R
- Getting started with `package.skeleton`
- In Rstudio: New Project -> New Directory -> R package


## Git

- Version control system originally for software development
- Useful for writing research articles, theses, R packages, ...
- Basics are easy and sufficient for small projects
- Embedded in Rstudio (Tools -> Version Control)
- http://rogerdudler.github.io/git-guide/

## Github

- Web-based Git repository hosting service
- Interacting with other developers and users
- Bug tracking, feature requests, task management, ... <!--assign tasks -->
- Free for public repositories
    + Free private repositories e.g. in [Bitbucket](https://bitbucket.org/)
- https://github.com/helske



## Useful packages for graphics

- `ggplot2`
    + Plotting system based on the grammar of graphics
    + Easy to produce complex multi-layered graphics
- `ggvis`
    + Similar to `ggplot2`, faster but more restricted <!-- newer, more features coming --> 
    + Interactive graphics in RStudio or a browser
- `grid`, `gridBase`
    + Control and flexibility in appearance and arrangement
    + Basis for developing high-level functions <!-- (e.g., lattice and ggplot2 packages -->
    <!-- grid cannot be used with base graphics, gridbase available for mixing grid and base graphics -->

## Useful packages for data manipulation

- `magrittr`
    + Piping via `%>%` operator
    + Improves readability and maintainability of code
- `dplyr` and `data.table`
    + Enhanced versions of `data.frame`
    + Fast and memory-efficient
    + More flexible data manipulation
    + Working with remote databases, automatic translation to SQL (`dplyr`)


## Useful packages for reporting

- `knitr`
    + Dynamic report generation (PDF, html, Word, ...)
    + Easy to re-compile and update outputs <!-- e.g. PDFs --> 
- `shiny`
    + Building interactive web applications from R
    + Web development skills not required


## Useful packages for packaging

- `testthat`
    + Testing framework for R
    + Catching errors, warnings, messages, ...  
    <!-- store tests so they can be re-run automatically when code gets changed --> 
    <!-- Easy to learn and use --> 
- `devtools`
    + Simplifying tasks in package development
- `roxygen2`
    + Writing documentation for functions etc.
- `Rcpp`
    + Simple C++ integration in R
    + Writing C++ separately, Rcpp handles the ugly stuff
<!-- - `inline`
    + Dynamically defining R functions with inlined C/C++/Fortran code --> 


## Learning

- Read
    + [RStudio Online learning](https://www.rstudio.com/resources/training/online-learning/)
    + [Quick-R](http://www.statmethods.net/)
    + [OpenIntro](https://www.openintro.org/)
- Hands-on learning online
    + [DataCamp](https://www.datacamp.com/)
        * R tutorials and data science courses in browser
        <!-- * Checks code on-the-fly, gives useful feedback -->
        <!-- * Registration needed, many courses are free One free month was enough for me -->
    + [Codeschool](https://www.codeschool.com/)
        * Learn R, Git, SQL, ...
        <!-- * Basics are free, some accessible without registration -->
- Hands-on in R
    + [`swirl` package](http://swirlstats.com/)

## Teaching I

- [OpenIntro](https://www.openintro.org/)
    + Free material for teachers to use and modify
    + Textbooks, slides, excercises, ...
    + Statistics, R, SAS, ...
- [DataCamp](https://www.datacamp.com/)
    + Resources for building own courses
- [`testwhat` package](https://github.com/Data-Camp/testwhat)
    + Wrapper around `testthat` for checking exercises
    

## Teaching II

- [Shiny apps](http://statistics.calpoly.edu/shiny) by Cal Poly State University
    + Correlation and regression game, sampling distribution demonstration, longest run of heads or tails, ...
- [Using Github](https://github.com/rundel/Presentations/blob/master/UseR2015/user2015.pdf) (experiences by Colin Rundel)
    + Assignments turned in via Github
    + Students learn version control
    + Simpler course administration
    + Keeping track of contributions
    <!-- He talks about automating checks -> could use checks for CRAN -->


## Getting help

- [Stack overflow](http://stackoverflow.com/)
- [Google](http://google.com/)...