Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Objects
Assign recognizable, concise names to your objects!
Let’s start with the object MyData - this can be anything. Let’s assume it is a
raster of NDVI values during the year 1982. How do we come up with a better
name?
Specificity in your naming helps you
keep track of your data in your code:
- NDVI1982
Classes
of objects determine how they
get handled. Add them as a suffix:
- NDVI1982_ras
The same goes for loop indicators! Avoid single-letter indicators and instead
use something like Iter_Years - an indicator iterating (Iter) over years.
Also, always use <- for assigning objects!
Aarhus University Biostatistics - Why? What? How? 4 / 18
Quality of Life Improvements Coding in R
Quality of Life Improvements
Use spaces to improve readability of your code:
c(1,2,rep(3,5)) # bad
c(1, 2, rep(3, 5)) # good
Insert line breaks to improve workflow. I like to break multiple lines of one
function call into logical blocks. For example, in ggplot2, the first line holds
data, the second contains aesthetics, the third adds a layer, and so on.
To avoid side-scrolling, use the automatic soft-wrap function in RStudio.
Be consistent in your coding style. Found a tidyverse solution to your
problem online but don’t use
tidyverse
syntax in the rest of your document?
Adapt the solution to be in the same style as the rest of your coding!
Aarhus University Biostatistics - Why? What? How? 5 / 18
Quality of Life Improvements Coding in R
Quality of Life Improvements
Use spaces to improve readability of your code:
c(1,2,rep(3,5)) # bad
c(1, 2, rep(3, 5)) # good
Insert line breaks to improve workflow. I like to break multiple lines of one
function call into logical blocks. For example, in ggplot2, the first line holds
data, the second contains aesthetics, the third adds a layer, and so on.
To avoid side-scrolling, use the automatic soft-wrap function in RStudio.
Be consistent in your coding style. Found a tidyverse solution to your
problem online but don’t use
tidyverse
syntax in the rest of your document?
Adapt the solution to be in the same style as the rest of your coding!
Aarhus University Biostatistics - Why? What? How? 5 / 18
Quality of Life Improvements Coding in R
Quality of Life Improvements
Use spaces to improve readability of your code:
c(1,2,rep(3,5)) # bad
c(1, 2, rep(3, 5)) # good
Insert line breaks to improve workflow. I like to break multiple lines of one
function call into logical blocks. For example, in ggplot2, the first line holds
data, the second contains aesthetics, the third adds a layer, and so on.
To avoid side-scrolling, use the automatic soft-wrap function in RStudio.
Be consistent in your coding style. Found a tidyverse solution to your
problem online but don’t use
tidyverse
syntax in the rest of your document?
Adapt the solution to be in the same style as the rest of your coding!
Aarhus University Biostatistics - Why? What? How? 5 / 18
Quality of Life Improvements Coding in R
Quality of Life Improvements
Use spaces to improve readability of your code:
c(1,2,rep(3,5)) # bad
c(1, 2, rep(3, 5)) # good
Insert line breaks to improve workflow. I like to break multiple lines of one
function call into logical blocks. For example, in ggplot2, the first line holds
data, the second contains aesthetics, the third adds a layer, and so on.
To avoid side-scrolling, use the automatic soft-wrap function in RStudio.
Be consistent in your coding style. Found a tidyverse solution to your
problem online but don’t use
tidyverse
syntax in the rest of your document?
Adapt the solution to be in the same style as the rest of your coding!
Aarhus University Biostatistics - Why? What? How? 5 / 18
Quality of Life Improvements Documenting Your Work
Comments
Don’t be overly proud of your coding skills. Comment everything!
Bad comments:
- ambiguous
- sparse
# Data Manipulation
Good comments:
- specify (what is the code doing)
- justify (why is it being done)
# Z-Score calculation for comparability
Write comments as though you were coding for someone else!
Aarhus University Biostatistics - Why? What? How? 6 / 18
Quality of Life Improvements Documenting Your Work
Comments
Don’t be overly proud of your coding skills. Comment everything!
Bad comments:
- ambiguous
- sparse
# Data Manipulation
Good comments:
- specify (what is the code doing)
- justify (why is it being done)
# Z-Score calculation for comparability
Write comments as though you were coding for someone else!
Aarhus University Biostatistics - Why? What? How? 6 / 18
Quality of Life Improvements Documenting Your Work
Comments
Don’t be overly proud of your coding skills. Comment everything!
Bad comments:
- ambiguous
- sparse
# Data Manipulation
Good comments:
- specify (what is the code doing)
- justify (why is it being done)
# Z-Score calculation for comparability
Write comments as though you were coding for someone else!
Aarhus University Biostatistics - Why? What? How? 6 / 18
Quality of Life Improvements Documenting Your Work
Comments
Don’t be overly proud of your coding skills. Comment everything!
Bad comments:
- ambiguous
- sparse
# Data Manipulation
Good comments:
- specify (what is the code doing)
- justify (why is it being done)
# Z-Score calculation for comparability
Write comments as though you were coding for someone else!
Aarhus University Biostatistics - Why? What? How? 6 / 18
Quality of Life Improvements Documenting Your Work
RMarkdown
Using Rmarkdown for your research comes with a multitude of advantages:
1 Entire workflow in one program (RStudio)
2 Great capabilities to present code
3 Research and reports reproducible at the click of one button
4 Combines R functionality and L
A
T
E
X formatting (if desired)
5 Consistent formatting
6 Clear presentation of code
7 Dynamic documents (you can generate various output document types)
8 Applicable for almost all document types you may desire as an output
(e.g. manuscripts, presentations, posters, etc.)
Aarhus University Biostatistics - Why? What? How? 7 / 18
Code Structure
Header
The Head is used as an information statement at the top of your code
document that informs the user of the contents, author, and (sometimes) date
of the last edit on said document:
# ################################################# #
# PROJECT: Project Title
# CONTENTS: What the code in your file is used for
# AUTHOR: Who worked on it
# EDIT: When the last edit was made
# ################################################# #
Personally, I don’t use the edit argument in my heads. Instead, I rely on version
control.
Aarhus University Biostatistics - Why? What? How? 9 / 18
Code Structure
Header
The Head is used as an information statement at the top of your code
document that informs the user of the contents, author, and (sometimes) date
of the last edit on said document:
# ################################################# #
# PROJECT: Project Title
# CONTENTS: What the code in your file is used for
# AUTHOR: Who worked on it
# EDIT: When the last edit was made
# ################################################# #
Personally, I don’t use the edit argument in my heads. Instead, I rely on version
control.
Aarhus University Biostatistics - Why? What? How? 9 / 18
Code Structure
Preamble
The Preamble is where you set up the most important parameters/guidelines
for your coding script. Personally, I highly recommend to make your first line in
the preamble read rm(list=ls()).
This is also where you load packages. I recommend doing so as follows:
install.load.package <- function(x) {
if (!require(x, character.only = TRUE))
install.packages(x, repos = "http://cran.us.r-project.org")
require(x, character.only = TRUE)
}
This functions checks whether the package is already installed. If it is, it will be
loaded. If it is not, it will be installed from CRAN and subsequently loaded.
You use it for multiple packages like this:
package_vec <- c("ggplot2", "foreach", "doParallel")
sapply(package_vec, install.load.package)
Aarhus University Biostatistics - Why? What? How? 10 / 18
Code Structure
Sections
Sections are a powerful tool in RStudio!
They keep your code:
- structured
- clear
- easy-to-use
They do so by:
- collapsing code
- adding a table of contents
→
It is a good idea to start an empty coding document by
pseudo-coding
your
analyses (i.e. writing sections of comments with what you want the code to do
and then do the actual coding).
Aarhus University Biostatistics - Why? What? How? 11 / 18
Code Structure
Sections
Sections are a powerful tool in RStudio!
They keep your code:
- structured
- clear
- easy-to-use
They do so by:
- collapsing code
- adding a table of contents
→
It is a good idea to start an empty coding document by
pseudo-coding
your
analyses (i.e. writing sections of comments with what you want the code to do
and then do the actual coding).
Aarhus University Biostatistics - Why? What? How? 11 / 18
Code Structure
Sections
Sections are a powerful tool in RStudio!
They keep your code:
- structured
- clear
- easy-to-use
They do so by:
- collapsing code
- adding a table of contents
→
It is a good idea to start an empty coding document by
pseudo-coding
your
analyses (i.e. writing sections of comments with what you want the code to do
and then do the actual coding).
Aarhus University Biostatistics - Why? What? How? 11 / 18
Code Structure
Sections
Sections are a powerful tool in RStudio!
They keep your code:
- structured
- clear
- easy-to-use
They do so by:
- collapsing code
- adding a table of contents
→
It is a good idea to start an empty coding document by
pseudo-coding
your
analyses (i.e. writing sections of comments with what you want the code to do
and then do the actual coding).
Aarhus University Biostatistics - Why? What? How? 11 / 18
Sharing Work Reproducibility
Reproducibility
Reproducibility is paramount in science.
Working directories
It is common to set a base directory at
the beginning of your code and refer to
sub-directories from there. We can do
better:
- getwd() in a project will yield the
directory of the project file
- the package here offers more
functions for this
- set paths to files or direcotires using
the file.path() function to avoid
cross-platform differences in folder path
syntax
Randomness
Random processes (e.g. sample(),
rnorm(), etc.) in R are always
pseudo-random (their calculation is
based on your computer time). We can
make sure the result of a random
process is always the same between
researchers by using:
- set.seed()
Aarhus University Biostatistics - Why? What? How? 13 / 18
Sharing Work Reproducibility
Reproducibility
Reproducibility is paramount in science.
Working directories
It is common to set a base directory at
the beginning of your code and refer to
sub-directories from there. We can do
better:
- getwd() in a project will yield the
directory of the project file
- the package here offers more
functions for this
- set paths to files or direcotires using
the file.path() function to avoid
cross-platform differences in folder path
syntax
Randomness
Random processes (e.g. sample(),
rnorm(), etc.) in R are always
pseudo-random (their calculation is
based on your computer time). We can
make sure the result of a random
process is always the same between
researchers by using:
- set.seed()
Aarhus University Biostatistics - Why? What? How? 13 / 18
Sharing Work Reproducibility
Reproducibility
Reproducibility is paramount in science.
Working directories
It is common to set a base directory at
the beginning of your code and refer to
sub-directories from there. We can do
better:
- getwd() in a project will yield the
directory of the project file
- the package here offers more
functions for this
- set paths to files or direcotires using
the file.path() function to avoid
cross-platform differences in folder path
syntax
Randomness
Random processes (e.g. sample(),
rnorm(), etc.) in R are always
pseudo-random (their calculation is
based on your computer time). We can
make sure the result of a random
process is always the same between
researchers by using:
- set.seed()
Aarhus University Biostatistics - Why? What? How? 13 / 18
Sharing Work Functions & Sourcing
Functions & Sourcing
Big projects lend themselves well to a multi-document workflow:
Functions
User-defined functions are a great way
of soft-coding your analyses to repeat
them for different input parameters and
I highly recommend doing so all the
time. Make sure your functions are:
- internally consistent
- well-documented
- easy to understand
Sourcing
Splitting your code into multiple
documents is a great idea to ensure
your project remains structured and
easy to handle. You can use the
source() function to combine those
documents. Keep in mind that you
should:
- use sensible file names
- sourced functions still need to be
called
Aarhus University Biostatistics - Why? What? How? 14 / 18
Sharing Work Functions & Sourcing
Functions & Sourcing
Big projects lend themselves well to a multi-document workflow:
Functions
User-defined functions are a great way
of soft-coding your analyses to repeat
them for different input parameters and
I highly recommend doing so all the
time. Make sure your functions are:
- internally consistent
- well-documented
- easy to understand
Sourcing
Splitting your code into multiple
documents is a great idea to ensure
your project remains structured and
easy to handle. You can use the
source() function to combine those
documents. Keep in mind that you
should:
- use sensible file names
- sourced functions still need to be
called
Aarhus University Biostatistics - Why? What? How? 14 / 18
Sharing Work Functions & Sourcing
Functions & Sourcing
Big projects lend themselves well to a multi-document workflow:
Functions
User-defined functions are a great way
of soft-coding your analyses to repeat
them for different input parameters and
I highly recommend doing so all the
time. Make sure your functions are:
- internally consistent
- well-documented
- easy to understand
Sourcing
Splitting your code into multiple
documents is a great idea to ensure
your project remains structured and
easy to handle. You can use the
source() function to combine those
documents. Keep in mind that you
should:
- use sensible file names
- sourced functions still need to be
called
Aarhus University Biostatistics - Why? What? How? 14 / 18
User-Friendliness Progress Bar
Progress Bar
A progress bar is a great way of keeping you updated on how your code is
progressing so far. It is especially useful when your code involves loops:
Data_vec <- 1:100 # a vector on integers from 1 to 100
# creating progress bar
ProgBar <- txtProgressBar(min = 0, max = length(Data_vec),
style = 3)
# looping over contents of Data_vec
for(Iter_ProgBar in 1:length(Data_vec)){
setTxtProgressBar(ProgBar, Iter_ProgBar) # updating ProgBar
} # end of Data_vec loop
Aarhus University Biostatistics - Why? What? How? 16 / 18
User-Friendliness Estimators
Estimators
Estimators are a great way to know when to come back to your computer or
server and check up on your data, code, and results. They are most useful in
loop based approaches as they need a baseline for the estimation:
Data_vec <- 1:100 # a vector on integers from 1 to 100
T_Begin <- Sys.time() # record time
# looping over contents of Data_vec
for(Iter_Est in 1:length(Data_vec)){
Sys.sleep(.1) # pause for .1 seconds
# estimator produced on first iteration
if(Iter_Est == 1){
T_End <- Sys.time() # record time
Duration <- as.numeric(T_End)-as.numeric(T_Begin) # time difference
print(paste("Estimated time to finish:",
as.POSIXlt(T_Begin + Duration
*
length(Data_vec),
tz = Sys.timezone(location=FALSE))
))
} # end of estimator check
} # end of Data_vec loop
## [1] "Estimated time to finish: 2020-03-25 00:33:50"
Yes, I did work on this presentation past midnight.
Aarhus University Biostatistics - Why? What? How? 17 / 18
User-Friendliness Parallel Processing
Parallel Processing
Got a big data approach where you carry out the same kind of analysis for
many different species or data sets? It’s taking forever to complete? Parallel
processing has your back! By default, R only uses one of the cores in your
computer or server for processing code. Use this, to make use of all the
computational power:
library(doParallel) # for registering clusters
library(foreach) # for parallel processing
Cores <- detectCores() # identify the number of cores in your machine
cl <- makeCluster(Cores) # create virtual cluster
registerDoParallel(cl) # register cluster of cores
# parallel processing
foreach(Iter_Par = 1:length(Data_vec)) %dopar% {
# your function here
} # end of parallel processing
stopCluster(cl) # stop cluster
Aarhus University Biostatistics - Why? What? How? 18 / 18
