Problem

plot(test_df[, 2:3],
     main  = "Testplot",
     lines(test_df[, 2:3])
points(test_df[1:3, 7:3],
       pch = 17,
       col = "redd",
       cex = 3)
text(test_df[, 2:3],
       labels = test_df[, 1],
       pos    = 4
       cex    = 1)

Answer

• lines can not be a parameter of plot
• index 7:3 is not defined
• color “redd”
• comma after the 4

plot(test_df[, 2:3],
     main  = "Testplot",
     lines(test_df[, 2:3])
points(test_df[1:3, 7:3],
       pch = 17,
       col = "redd",
       cex = 3)
text(test_df[, 2:3],
       labels = test_df[, 1],
       pos    = 4
       cex    = 1)

plot(test_df[, 2:3],
     main  = "Testplot")
lines(test_df[, 2:3])
points(test_df[1:3, 2:3],
       pch = 17,
       col = "red",
       cex = 3)
text(test_df[, 2:3],
       labels = test_df[, 1],
       pos    = 4,
       cex    = 1)

What is control flow?

Control flow

• Loops repeat code
  • for() certain number of repetitions
  • while() as long as a condition is met
  • repeat() until we break the loop
• Conditions run code if a condition is met
  • if() run some code or not
  • if() else() run some code or another code
• simple examples (both would be solved differently in practice)
  • calculate the mean artefact size for all sites
  • calculate the mean size for complete artefacts only

Control flow

For

For loops repeat code a certain times and use an internal iteration variable.

for (i in 1:5){
    print(paste("Loop no.: ", i))
}

## [1] "Loop no.:  1"
## [1] "Loop no.:  2"
## [1] "Loop no.:  3"
## [1] "Loop no.:  4"
## [1] "Loop no.:  5"

While

While loops repeat code until a certain condition is met. Iteration variables, if required have to be installed manually.

Analyze Code!

i <- 1
while(i < 5){
    print(paste("Loop no.: ", i))
    i <- i + 1
}

## [1] "Loop no.:  1"
## [1] "Loop no.:  2"
## [1] "Loop no.:  3"
## [1] "Loop no.:  4"

While

Iteration variable and the variable for the condition need not to be the same.

i <- 1
a <- 0
while(a < 5000){
    a <- 2^i
    print(paste("Loop no.: ", i))
    i <- i + 1
}

## [1] "Loop no.:  1"
## [1] "Loop no.:  2"
## [1] "Loop no.:  3"
## [1] "Loop no.:  4"
## [1] "Loop no.:  5"
## [1] "Loop no.:  6"
## [1] "Loop no.:  7"
## [1] "Loop no.:  8"
## [1] "Loop no.:  9"
## [1] "Loop no.:  10"
## [1] "Loop no.:  11"
## [1] "Loop no.:  12"
## [1] "Loop no.:  13"

If

If allows for conditional code. The condition contains a logical value and can make use of logical operators: ==, !=, <, <=, >, >=

The terms can be combined with and & and or |.

if(3 == 4){print("abc")}

Repeat

Repeat repeats code until we break the loop with break.

i <- 1
repeat{print(paste("Loop no.: ", i))
    i <- i + 1
    if(i < 5){next}
    print(paste("Value: ", 2^i))
    if(i > 10){break}
    }

## [1] "Loop no.:  1"
## [1] "Loop no.:  2"
## [1] "Loop no.:  3"
## [1] "Loop no.:  4"
## [1] "Value:  32"
## [1] "Loop no.:  5"
## [1] "Value:  64"
## [1] "Loop no.:  6"
## [1] "Value:  128"
## [1] "Loop no.:  7"
## [1] "Value:  256"
## [1] "Loop no.:  8"
## [1] "Value:  512"
## [1] "Loop no.:  9"
## [1] "Value:  1024"
## [1] "Loop no.:  10"
## [1] "Value:  2048"

If Else

If Else tests for a condition. If the condition is True Code 1 is terminated if False execute Code 2.

if(3 == 4){print("abc")
    }else{print("deff")}

## [1] "deff"

Conditions

• Loops are supposed to be slower and less elegant as vector based methods. In practice it is only relevant if they work and are fast enough for your purpose.

Write your own functions

Functions are container, shortcuts or names for pieces of code. Variables can be passed on to the functions as parameters.

analyze Code!

add <- function(a, b){
    c <- a + b
    return(c)
}

add(3, 5)

## [1] 8

Vector based methods: using vectors

x <- c(2, 4, 1, 5)
sqrt(x)

## [1] 1.414214 2.000000 1.000000 2.236068

mean(x)

## [1] 3

Vector based methods: apply and co.

apply is a function that runs other functions for every column or row of a matrix or dataframe. apply is usually faster than a loop.

analyze Code!

df <- data.frame(a = c(1, 2, 3, 4, 5),
                 b = c(5, 4, 3, 2, 1),
                 c = c(3, 5, 3, 7, 5))
apply(df,
      1,         # 1 for row; or 2 for column indexing
      mean)

## [1] 3.000000 3.666667 3.000000 4.333333 3.666667

apply(df,
      2,
      mean)

##   a   b   c 
## 3.0 3.0 4.6

Which approach do you prefer?

• Loop
• Apply

Tidyverse

Tidyverse is a philosophy and a style of data sciences within the R ecosphere, initiated by Hadley Wickham (now at RStudio). Tidyverse includes R-packages partly as part of the meta-package tidyverse. The following slides are mainly based on Wickham/Grolemund (2017): http://r4ds.had.co.nz/.

• Data concept
• Packages for data handling
• Programming style

Tidyverse

• Tidyverse-Packages:
  • ggplot2
  • tibble
  • tidyr
  • readr
  • purrr
  • dplyr
  • stringr
  • forcats
  • magrittr
  • modelr
  • glue
  • broom

library("tidyverse")

Tidy data

Tidy data

Tidy data

Tidy data

Tidy data

Readr

readr (Wickham/Hester 2021) reads data into tibbles.

library(readr)
ceramics <- readr::read_delim("../2data/22archdata/keramik.csv",
    ";",
    escape_double = FALSE,
    trim_ws = TRUE)
head(ceramics)

Tibbles

• replacement for dataframes
• no factors as default
• no rownames
• slicing tibbles always results in tibbles, even single numbers
  • pull() can be used to simplify tibbles

Pipe & Dplyr

The pipe (Bache/Wickham 2020) is forwarding the output of a function to the next function.

library(magrittr)
ceramics %>%
    dplyr::filter(Breite < 20) %>%
    dplyr::select(id,
                  Hoehe,
                  Breite,
                  Motive) -> ceramics2
head(keramik2)

Pipe & Dplyr (old approach)

library(magrittr)
ceramics2 <- ceramics %>%
    dplyr::filter(Breite < 20) %>%
    dplyr::select(id,
                  Hoehe,
                  Breite,
                  Motive)

Pipe & Dplyr

ceramics %>%
    dplyr::mutate(g = Hoehe > 10) %>%
    dplyr::rename(Groesse = g) %>%
    dplyr::group_by(Groesse) %>%
    dplyr::summarise(m = mean(Breite))

• poorman is a replacement for dplyr without tidyverse-integration.
• since R4.1, R has a native Pipe |>, which does not depend on a package

Tidyr

(Wickham 2021)

library(tidyr)
ceramics %>%
    tidyr::separate(Motive,
                    into = c("m1", "m2", "m3", "m4", "m5", "m6", "m7", "m8"),
                    sep = ",") -> ceramics3
head(ceramics3)

Tidyr

Analyze Code!

ceramics3 %>%
    tidyr::gather(key   = "key",
                  value = "value",
                  -id) -> ceramics4
head(ceramics4)

Tidyr

ceramics4 %>%
    tidyr::spread(key     = "key",
                  value   = "value",
                  convert = T) -> ceramics5
head(ceramics5)

ggplot2

ggplot2 (Wickham 2016) makes plots according to the “Grammar of Graphics”: data + geometry = graphic

• plot Object
• geom Object (geometry)
• aes Object (aesthetics)
• Layers

library(ggplot2)
ceramics3 %>%
    ggplot2::ggplot() +
    geom_point(mapping = aes(x     = Breite,
                             y     = Hoehe,
                             color = Dat))

ggplot2

plots can be assigned to an object

library(ggplot2)
ceramics3 %>%
    ggplot2::ggplot() -> p

ggplot2

  p +
    geom_density(mapping = aes(x = Breite)) +
    geom_point(mapping   = aes(x = Breite,
                               y = Hoehe,
                               color = Dat)) +
    geom_dotplot(mapping = aes(x = Breite)) +
    ggtitle("Braubacher Stempelkeramik") +
    theme_minimal()

ggplot2

library(ggplot2)
ceramics %>%
    tidyr::separate(Motive,
                    into = c("m1", "m2", "m3", "m4", "m5", "m6", "m7", "m8"),
                    sep = ",") %>%
    dplyr::mutate(g = Hoehe > 10) %>%
    ggplot2::ggplot() -> p2

    p2 + geom_point(mapping = aes(x = Breite,
                                  y = Hoehe,
                                  color = g,
                                  size  = 4,
                                  shape = m3)) +
    facet_wrap( ~ Dat)

Exercise

• Load data from archdata and calculate some mean and sums values.

• calculate the sum of all numbers from 15 to 75 by

  • using the sum() function and
  • a loop

• Write a function, which divides a number a by a number b.

• Calculate all length by high values (ore similar numbers) for the fibulae from archdata by

  • using apply(),
  • a simple vector calculation and
  • a loop

• write a loop converting the long-form data to wide-form

  • df <- data.frame( key = c(“a”, “a”, “b”, “a”, “b”, “b”, “c”, “c”, “c”), value = c(1, 3, 54, 2, 66, 43, 743, 234, 444) )
  • target: df2 <- data.frame( a = c(), b = c(), c = c() )

We are done for today!