Skip to content

Latest commit

 

History

History
305 lines (243 loc) · 8.25 KB

IntroToR.md

File metadata and controls

305 lines (243 loc) · 8.25 KB

Introduction To R

author: Kevin Shook date: November 23, 2017 autosize: true css: style.css

Objectives

  • To explain what R is, and what it can be used for
    • Will focus on why and what, rather than how
    • Future seminars will cover details of how to use R
    • Will be giving a live demonstration of some of the capabilities of R

Typical research workflow:

  • Reading in data (text files, databases, .xls)
  • Data massaging
  • Data exploration (trial calculations, plotting)
  • Final calculations
  • Saving results
  • Exporting data for other programs to use
  • Creating publication graphs
  • Writing a paper/thesis

Reproducible research

  • Need to know what you did, and to be able to re-do it
  • Have to justify your results
  • Need to be able to re-do work due to changes or mistakes

What is R?

  • R began as a statistical programming language
    • It's now a general-purpose scientific program
    • R allows you to write scripts to automate your work
  • Can combine text, equations, R code, output and figures in a single output document
    • Creates automatically-updated documents
    • Results in self-documenting, reproducible research

Why "R"?

  • S-plus is a proprietary statistics program
    • uses the S language
  • R is a Free Open Source implementation of the S language

Why use R?

  • Excellent for statistics, data manipulation and graphing
  • Free Open Source Software
    • Can see, test and verify the source code
  • Uses standard file formats - no lock-in
  • Huge number of packages available
  • Works well with other programs

Statistics

  • R is the standard program for statistical analyses
  • Widely used for teaching statistics
  • Can do any type of statistical analyses that you need

Data crunching

  • R is excellent for massaging for all types of scientific data
    • can read data from almost any source including spreadsheets and databases
    • time series
    • spatial data
    • categorical data
  • Widely used for "big" data

Graphing

  • R is arguably the best program for scientific graphing


GIS

  • R can do very sophisticated GIS analyses

Getting R

Packages

  • Enormous amount of R-code is available

R demonstration

type: prompt

Graphing

  • Standard (built-in) graphing uses the command plot:
plot(xvals, yvals, options)
  • Easy to use from the command line
  • Good for quick and dirty plots
  • Can get better results for publication using another package
plot(c(1,2,3), c(4,5,6), type="p", col="red", cex=2, pch=19)

plot of chunk unnamed-chunk-2

ggplot2

Why ggplot2?

  • Creates amazing publication-quality graphics very easily
  • Based on work of Edward Tufte
  • Uses a grammar for graphs
  • Can change graphs interactively
  • Extremely good for categorized data

Grammar of graphing

  • Graphs are made of
Definition Short name
Aesthetics aes
Geometric objects geom
Statistical transformations stat
Scales scale
Faceting facet
Theme theme

Creating a ggplot2 graph

class: small-code

  • Create a ggplot2 object in a variable
p <- ggplot(dataframe)
  • Add an aesthetic defining the columns
p <- p + aes(xvals, yvals)
  • Add a geometry
p <- p + geom_point()
  • Add stats, themes, scales, facets
p <- p + theme_gray(18) +
xlim(0, 5)
  • Display - type the variable name
p
  • Save to a file
ggsave("graphfile.png")

ggplot2 data

  • ggplot2 requires values to be stored in data frames that are tall, not wide
  • Opposite of standard R graphs
    • Takes some getting used to
  • Worth the effort, as it is much more powerful
  • Allows you to use categories in your plots
  • Tools available to convert your data from wide to tall

Wide data

  • Like a spreadsheet: each variable's value in a separate column
  • Inflexible, doesn't allow for multiple classifications
  • Doesn't deal well with differing numbers of values
  • Doesn't tell us what the data represents
    • not very reproducible
Time Saskatoon Regina Calgary
00:00:00 -7 -7 -1
01:00:00 -5 -9 -2
02:00:00 -5 -9 -3
03:00:00 -6 -1 -2
04:00:00 -6 -9 -3
05:00:00 -6 -11 NA

Tall data

Time Temp Location
00:00:00 -7 Saskatoon
01:00:00 -5 Saskatoon
02:00:00 -5 Saskatoon
03:00:00 -6 Saskatoon
04:00:00 -6 Saskatoon
05:00:00 -6 Saskatoon
00:00:00 -7 Regina
01:00:00 -9 Regina
02:00:00 -9 Regina
03:00:00 -1 Regina
04:00:00 -9 Regina
05:00:00 -11 Regina
00:00:00 -1 Calgary
...

ggplot2 demonstration

type: prompt

Challenges

  • steep learning curve
    • have to learn many new commands

"R makes easy things hard and hard things easy"

But:

  • lots of support and information available
  • will be doing more training

Resources

Rseek (Google for R):
http://rseek.org/

R reference card:
https://cran.r-project.org/doc/contrib/Baggott-refcard-v2.pdf

Books and manuals:
An Introduction to R
https://cran.r-project.org/doc/manuals/r-release/R-intro.pdf

R for beginners
https://cran.r-project.org/doc/contrib/Paradis-rdebuts_en.pdf

The R guide
https://cran.r-project.org/doc/contrib/Owen-TheRGuide.pdf

The R Reference Index:
https://cran.r-project.org/doc/manuals/r-release/fullrefman.pdf

Centre for Hydrology R packages

  • There are several R packages developed for accessing/processing data
package functions
CRHMr pre- and post- processing for CRHM
MSCr reads MSC data
Reanalysis reads gridded reanalysis data
WISKIr reads from WISKI database
HYDAT reads WSC HYDAT data

This presentation