Environment, Packages, Functions, Import/Export

Install packages… and more

Traditional installing of a package from CRAN:

install.packages("effsize")

installing of multiple packages from CRAN at once:

install.packages( c("effsize","psych","ggplot2") )

For development or personal use, you may occasionally install packages from outside CRAN, such as from GitHub:

devtools::install_github("FilippoGambarota/filor")
devtools::install_github("EnricoToffalini/toffee")

After installing, you need to load the packages using function library:

library(effsize)
library(ggplot2)

Install packages… and more

call functions

After loading a package, its functions are directly callable throughout the R session:

library(psych)

fisherz(rho=0.8) # use a function from the "psych" package

[1] 1.098612

you may directly call any function from any installed package, even without loading it, using “::”; this is especially useful when there is a risk of functions with conflicting names, or if you just don’t want to load an entire package for using a single function:

psych::fisherz(rho=0.8)

[1] 1.098612

Functions and arguments

Functions typically take some input parameters, known as arguments, process that, and yield some output/result(s)

Functions and arguments

arguments

• values or variables you pass to a function as input, or to control its behavior

for example, seq() generates a sequence of numbers; “from” and “to” are arguments: it will provide the integers between these two extremes:

seq(from = 3, to = 7)

[1] 3 4 5 6 7

length.out controls how many equally spaced numbers must be generated:

seq(from = 3, to = 7, length.out = 4)

[1] 3.000000 4.333333 5.666667 7.000000

alternatively, by defines the step size between numbers:

seq(from = 3, to = 7, by = 0.6)

[1] 3.0 3.6 4.2 4.8 5.4 6.0 6.6

Functions and arguments

arguments

• values or variables you pass to a function as input, or to control its behavior

rnorm() will generate “n” random numbers from a normal distribution with “mean” as the average and “sd” as the standard deviation:

rnorm(n = 5, mean = 100, sd = 15)

[1] 106.79681  70.22668  99.49541  74.45967 114.30017

Positional matching - know that arguments names may be omitted if placed in the correct order

rnorm(5, 100, 15)

[1]  92.09907  82.05890  67.75271 102.77268  54.38091

Functions and arguments

Default arguments - a function might still work even if some arguments are omitted, as it can use its own default values (in this case “mean=0, sd=1”)

rnorm(n = 5)

[1]  0.5395298 -0.2850636 -1.7526744 -0.9979548 -0.4308542

Errors - however, omitting mandatory arguments will result in an Error

rnorm(mean = 100, sd = 15)

Error in rnorm(mean = 100, sd = 15): argument "n" is missing, with no default

Warnings - Some inputs may cause the function to produce Warnings and bad output, but do not stop code execution

rnorm(n = 5, mean = 100, sd = -15)

Warning in rnorm(n = 5, mean = 100, sd = -15): NAs produced

[1] NaN NaN NaN NaN NaN

Functions and arguments

HELP! see the documentation of a function

There are two ways to access documentation: using “?” and using help()

?rnorm # this will work
help(rnorm) # this does the same

Set up Working Directory

getwd()

[1] "C:/Users/enric/OneDrive - Università degli Studi di Padova/Documents/_DIDATTICA/R for Data Science/_GITHUB Basics R DataScience/Slides"

Set up Working Directory

As a general rule:

• When you open R or the RStudio app, the default WD may be the documents folder (in Windows) or the home directory (e.g., /home/username; in Linux or macOS);

• This default may be reset at any time from inside RStudio on Tools > Global Options... > General;

• When RStudio is newly open by opening a file (e.g., a .R script file), the WD may be set at that file location (actually my favorite);

• However, you can set a new WD at any time from within the R code, using the setwd() function, for example:

setwd( "C:/Users/enric/" )

Set up Working Directory

Set up Working Directory

Finally, not vital for now, but know the difference between:

• Absolute path: "C:/Users/enric/" indicates the full directory path from the root

• Relative paths: for import/export purposes you may move around the current WD

  • for example png(filename="figures/Fig1.png") may save \(Fig1.png\) into the \(figures\) directory which is inside the current WD;
  • differently, png(filename="../figures/Fig1.png") may save \(Fig1.png\) into the \(figures\) directory which is outside, one level up the current WD

Import/export

Import/export

Workspace

All your R code (script) is generally stored in text files with a .R extension. But where do you save your results and objects?!

You can export the entire workspace (with all your objects) using the save.image() function:

# let's populate the workspace first
myName = "Enrico"
prof = TRUE
coursesTaught = 4L
age = 36

# now let's save it
save.image("myWS.RData") 

Specifying "myWS.RData" is not mandatory but recommended, otherwise your file will simply be named ".RData". (By the way… where will it be saved?)

Import/export

Workspace

Alternatively, you may even save just one or a few workspace objects, rather than all:

# let's populate the workspace first
myName = "Enrico"
prof = TRUE
coursesTaught = 4L
age = 36

# now let's save only two objects
save(myName, age, file="myWS.RData") 

This will save only variables myName and age into a newly created file named myWS.RData

This may be useful when you have an overcrowded workspace and prefer to save only a few objects that store the final results

Import/export

Workspace

Once you open a new R session, you may load the previously stored workspace using the load() function, specifying load("workspace_name.RData"), like this:

# empty the workspace to make sure there's actually nothing!
rm(list=ls())
ls()

# now load the previously saved workspace
load("myWS.RData") 
# make sure that the objects have been loaded
ls()

Import/export

Data

Import/export

Data

Here’s an example of using the read.csv() function for importing data:

# IMPORT csv data from a "data" subfolder, and store it in an object named "df"
df = read.csv("data/Performance.csv", header=TRUE, sep=",", dec=".")

head(df) # have a look at the first few rows

  id  name anx acc     time
1  1 nydga  20  15 2.077932
2  2 bwknr  14   9 2.436858
3  3 sauuj  18  12 2.549814
4  4 vnjgi  27  15 4.386718
5  5 oueiy  21  11 5.248933
6  6 neebj  12  13 3.463094

Import/export

Data

If you absolutely want to import your data directly from a MS Excel document (.xlsx), you may use function read_excel() from the package readxl:

library(readxl)
df = data.frame( read_excel("data/Performance.xlsx") )
# data.frame() forces it to be a dataframe, otherwise it's a tibble
head(df)

  id  name anx acc     time
1  1 nydga  20  15 2.077932
2  2 bwknr  14   9 2.436858
3  3 sauuj  18  12 2.549814
4  4 vnjgi  27  15 4.386718
5  5 oueiy  21  11 5.248933
6  6 neebj  12  13 3.463094

Import/export

Data

A good trick if you don’t want to specify any relative or absolute path, and want to manually select data each time, is using the file.choose() function:

df = read.csv(file.choose(), header=TRUE, sep=",", dec=".")

Import/export

Data

Other “tricks” for importing data involve using the functions in the RStudio menu, particularly:

• File > Import Dataset > From text (base)…

• File > Import Dataset > From Excel

• File > Import Dataset > From SPSS…

However … using these functions is not best practice, because they are specific to the RStudio IDE. It’s better to use code for reproducibility

Import/export

Data

You have processed data with R, now… how to export it?

When collaborating with someone also using R, you may choose to exchange data directly by exporting the object or the entire workspace as a .RData file, using the save() or save.image() function respectively.

However, if you need to export your data in a more universally readable tabular format, such as .csv, you may use write.table():

# specify the dataframe to export (here named "df")
# along with the desired file name, and other arguments

write.table(df, file="myExportedData.csv", sep=",", row.names=F)

Import/export

Figures

R has a collection of functions for exporting figures in different formats: pdf(), png(), jpeg(), bmp(), tiff(), svg().

Here is an example using png() :

# set up a graphic output file named "MyFigure.png" with some settings
png("MyFigure.png", height=1500, width=2000, units="px", res=300)

# code for creating a simple boxplot
boxplot(iris$Sepal.Width) 

# close the graphic output file and actually export the plot
dev.off()