Data Visualisation with R Workshop

Slides for the introduction part of the "Data Visualisation with R" Workshop at Applied Machine Learning Days 2020

Transcript

1. Applied Data Science Data visualisation and communication Xavier Adam @xvrdm Workshop Introduction
2. Applied Data Science Data visualisation and communication
3. THE LAST MILE
7. About the COS program
8. Applied Data Science Data visualisation and communication 1 2 3
9. Theory P Source: Dark Horse Analytics
10. Theory P NOT GREAT BETTER Source: Dark Horse Analytics
11. Theory P Source: Dark Horse Analytics
12. Practice
13. Theory Practice
14. *Identification with actual EPFL professors (living or deceased), is intended and should be inferred.
15. Let’s lift 90 KILOGRAMS !
16. Let’s import 90 GIGABYTES !
17. Applied Data Science Data visualisation And (data) communication 1 2 3
18. Import Tidy Transform Visualise Model Communicate Full Spectrum Data Science: A workflow
19. Import Tidy Transform Visualise Model Communicate Visualisation Data Science: A workflow ~FANTASY ~
20. Import Tidy Transform Visualise Model Communicate Visualisation Data Science: A workflow ~REALITY ~
21. Applied Data Science Data visualisation And (data) communication Show practical ways to share visualisation Show practical ways to produce visualisation Show practical ways to share visualisation
22. About the workshop
23. Today’s Menu
24. Today’s Menu Visualisation for time series Visualisation for geospatial data Visualisation for network data Common thread: Interactivity through webapps ~NOW ~ Brief overview of the R syntax
25. It isn’t a comprehensive workshop about visualisation in R, let alone the R ecosystem… WHAT THIS WORKSHOP ISN’T
26. It isn’t a comprehensive workshop about {ggplot2}, let alone the R ecosystem… WHAT THIS WORKSHOP ISN’T How to draw an Owl 2. Draw the entire damn owl A beginners-friendly workshop by the Extension School 1. Start with 
 two simple circles
27. WHAT THIS WORKSHOP ISN’T … nor a 101 tutorial about data visualisation in R
28. THIS IS SIMPLY A BIRD EYE VIEW Let’s peek under the hood of modern visualisation in R
29. What this workshop is: A big plate of appetizers
33. Simple vs Easy Complex vs Difficult
34. Why would you even consider using code for visualisation? WHY BOTHER?
35. Endless possibilities Unlocked Complexity WHY BOTHER? Full reproducibility
36. WHY BOTHER? Time to create Number of times Full reproducibility Code Point and Click 1 2 3 4
37. WHY BOTHER? Time to create Number of times Full reproducibility 1 2 3 4 Time to improve
38. WHY BOTHER? Time to create Visualisation Complexity Unlocked complexity Point and Click
39. WHY BOTHER? Time to create Visualisation Complexity Code Point and Click Unlocked complexity
40. WHY BOTHER? Time to create Visualisation Complexity Code Point and Click Unlocked complexity TABLE-AU POWERB-I - -
41. WHY BOTHER? Time to create Visualisation Complexity Code Point and Click Unlocked complexity TABLE-AU POWERB-I - -
42. WHY BOTHER? Complexity Code ity TABLE-AU POWERB-I -MY GOD -NEED TO GET OUT OF HERE!
43. Endless possibilities WHY BOTHER? Sources: @diegovalle / @hrbrmstr / @lenkiefer / Lea Waniek
44. WHY BOTHER? Endless possibilities Sources: @hrbrmstr / @committedtotape
45. WHY BOTHER? Endless possibilities Sources: @katiejolly6 / @Christi58451746 / @DataPlanes
46. 10’000 foot-view() of the modern R ecosystem IS IT FOR ME?
47. In the pilot seat
48. In the pilot seat Where you write code
49. In the pilot seat Where you write code ===== Where you see the output
50. Communication through output IS IT FOR ME?
51. In the pilot seat Where you write code ===== Where you see the output
52. In the pilot seat ===== Where you see the output
53. In the pilot seat ===== Where you see the output
54. In the pilot seat ===== Where you see the output
55. In the pilot seat ===== Where you see the output
56. In the pilot seat ===== Where you see the output
57. ===== Where you see the output In the pilot seat
58. Visualisation through expressive code IS IT FOR ME?
59. In the pilot seat Where you write code ===== Where you see the output
60. your very own data-working bench
61. PACKAGES ARE HERE TO HELP
62. AND THEY WORK WELL TOGETHER
63. Changing data with verbs (writing data wrangling sentences) > 110110010101010100 > data, please do this! > 1110010101101011101
64. AN EXAMPLE WITH DPLYR A R package giving you a Toolbox for common data manipulation
65. How to keep only certain columns in a table? ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL SIZE AGE DOG 80 9 CAT 40 12 BIRD 20 2 DOG 70 7 data
66. data[c(“ANIMAL”, “AGE”)] ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL SIZE AGE DOG 80 9 CAT 40 12 BIRD 20 2 DOG 70 7 data select() Only the columns you need, in the order you want Before
67. ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 Data [ c (“ANIMAL”, “AGE”) ] ANIMAL SIZE AGE DOG 80 9 CAT 40 12 BIRD 20 2 DOG 70 7 data select() Only the columns you need, in the order you want Before
68. ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 Data [ c (“ANIMAL”, “AGE”) ] ANIMAL C2 AGE DOG 9 CAT 12 BIRD 2 DOG 7 df select() Only the columns you need, in the order you want Before
69. select(data, ANIMAL, AGE) select() Only the columns you need, in the order you want After data[c(“ANIMAL”, “AGE”)] ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL SIZE AGE DOG 80 9 CAT 40 12 BIRD 20 2 DOG 70 7 data
70. How to keep only certain rows in a table? ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL AGE DOG 9 DOG 7
71. data[data$ANIMAL == “DOG”,] filter() Only the rows you need ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL AGE DOG 9 DOG 7 Before
72. filter(data, ANIMAL == “DOG”) filter() Only the rows you need data[data$ANIMAL == “DOG”,] ANIMAL AGE DOG 9 CAT 12 BIRD 2 DOG 7 ANIMAL AGE DOG 9 DOG 7 After
73. How to rename columns in a table? ANIMAL AGE DOG 9 DOG 7 BREED AGE DOG 9 DOG 7
74. rename() Columns with the names you need names(data) <- c(“BREED”, “AGE”) ANIMAL AGE DOG 9 DOG 7 BREED AGE DOG 9 DOG 7 Before
75. rename(data, BREED = ANIMAL) rename() Columns with the names you need names(data) <- c(“BREED”, “AGE”) ANIMAL AGE DOG 9 DOG 7 BREED AGE DOG 9 DOG 7 After
76. R Base Syntax (only option back then) R Tidyverse Syntax (possible today) data[c(“ANIMAL”, “AGE”)] select(data, ANIMAL, AGE) data[data$ANIMAL == “DOG”,] filter(data, ANIMAL == “DOG”) names(data) <- c(“BREED”, “AGE”) rename(data, BREED = ANIMAL)
77. R Base Syntax (only option back then) R Tidyverse Syntax (possible today) data[c(“ANIMAL”, “AGE”)] select(data, ANIMAL, AGE) data[data$ANIMAL == “DOG”,] filter(data, ANIMAL == “DOG”) names(data) <- c(“BREED”, “AGE”) rename(data, BREED = ANIMAL)
78. SELECT ! FILTER ! RENAME ! FULL DATA KET R Tidyverse Syntax (possible today) select(data, ANIMAL, AGE) filter(data, ANIMAL == “DOG”) rename(data, BREED = ANIMAL)
79. psss… select ! psss… filter ! pss… rename ! DATA R Tidyverse Syntax (possible today) select(data, ANIMAL, AGE) filter(data, ANIMAL == “DOG”) rename(data, BREED = ANIMAL)
80. R Base Syntax (only option back then) R Tidyverse Syntax (possible today) data[c(“ANIMAL”, “AGE”)] select(data, ANIMAL, AGE) data[data$ANIMAL == “DOG”,] filter(data, ANIMAL == “DOG”) names(data) <- c(“BREED”, “AGE”) rename(data, BREED = ANIMAL)
81. R Base Syntax (only option back then) R Tidyverse Syntax (possible today) data[c(“ANIMAL”, “AGE”)] select(data, ANIMAL, AGE) data[data$ANIMAL == “DOG”,] filter(data, ANIMAL == “DOG”) names(data) <- c(“BREED”, “AGE”) rename(data, BREED = ANIMAL)
82. Changing data with verbs (writing data wrangling sentences) > 110110010101010100 > data, please do this! > 1110010101101011101
83. Combining sentences into paragraphs > data, please do this! And this And also this
84. data1 <- select(data, ANIMAL, AGE) data2 <- filter(data1, ANIMAL == “DOG”) data3 <- rename(data3, BREED = ANIMAL)
85. data1 <- select(data, ANIMAL, AGE) data2 <- filter(data1, ANIMAL == “DOG”) data3 <- rename(data3, BREED = ANIMAL)
86. ENTER THE PIPE SIGN (I.E. “THEN”) %>%
87. NOT THIS PIPE! THE OTHER PIPE!
88. NOT THIS PIPE! THE OTHER PIPE! %>% %>% %>% data
89. data1 <- select(data, ANIMAL, AGE) data2 <- filter(data1, ANIMAL == “DOG”) data3 <- rename(data2, BREED = ANIMAL)
90. data1 <- select(data, ANIMAL, AGE) %>% data2 <- filter(data1, ANIMAL == “DOG”) %>% data3 <- rename(data2, BREED = ANIMAL)
91. data1 <- select(data, ANIMAL, AGE) %>% data2 <- filter(data1, ANIMAL == “DOG”) %>% data3 <- rename(data2, BREED = ANIMAL)
92. select(data, ANIMAL, AGE) %>% filter(ANIMAL == “DOG”) %>% rename(BREED = ANIMAL)
93. select(data, ANIMAL, AGE) %>% filter(ANIMAL == “DOG”) %>% rename(BREED = ANIMAL)
94. data %>% select(ANIMAL, AGE) %>% filter(ANIMAL == “DOG”) %>% rename(BREED = ANIMAL)
95. data %>% select(ANIMAL, AGE) %>% filter(ANIMAL == “DOG”) %>% rename(BREED = ANIMAL) A script by any other name…
96. `dplyr!::add_count()` `dplyr!::add_count_()` `dplyr!::add_row()` `dplyr!::add_rownames()` `dplyr!::add_tally()` `dplyr!::add_tally_()` `dplyr!::all_equal()` `dplyr!::all_vars()` `dplyr!::anti_join()` `dplyr!::any_vars()` `dplyr!::arrange()` `dplyr!::arrange_()` `dplyr!::arrange_all()` `dplyr!::arrange_at()` `dplyr!::arrange_if()` `dplyr!::as_data_frame()` `dplyr!::as_label()` `dplyr!::as_tibble()` `dplyr!::as.tbl()` `dplyr!::as.tbl_cube()` `dplyr!::auto_copy()` `dplyr!::band_instruments()` `dplyr!::band_instruments2()` `dplyr!::band_members()` `tidyr!::complete()` `tidyr!::complete_()` `tidyr!::crossing()` `tidyr!::crossing_()` `tidyr!::drop_na()` `tidyr!::drop_na_()` `tidyr!::expand()` `tidyr!::expand_()` `tidyr!::extract()` `tidyr!::extract_()` `tidyr!::extract_numeric()` `tidyr!::fill()` `tidyr!::fill_()` `tidyr!::full_seq()` `tidyr!::gather()` `tidyr!::gather_()` `tidyr!::nest()` `tidyr!::nest_()` `tidyr!::nesting()` `tidyr!::nesting_()` `tidyr!::population()` `tidyr!::replace_na()` `tidyr!::separate()` `tidyr!::separate_()` `tidyr!::separate_rows()` `lubridate!::as_datetime()` `lubridate!::as.difftime()` `lubridate!::as.duration()` `lubridate!::as.interval()` `lubridate!::as.period()` `lubridate!::ceiling_date()` `lubridate!::Compare()` `lubridate!::date()` `lubridate!::date_decimal()` `lubridate!::date!<-()` `lubridate!::day()` `lubridate!::day!<-()` `lubridate!::days()` `lubridate!::days_in_month()` `lubridate!::ddays()` `lubridate!::decimal_date()` `lubridate!::dhours()` `lubridate!::dmicroseconds()` `lubridate!::dmilliseconds()` `lubridate!::dminutes()` `lubridate!::dmy()` `lubridate!::dmy_h()` `lubridate!::dmy_hm()` `lubridate!::dmy_hms()` `lubridate!::dnanoseconds()` `stringr!::boundary()` `stringr!::coll()` `stringr!::fixed()` `stringr!::fruit()` `stringr!::invert_match()` `stringr!::regex()` `stringr!::sentences()` `stringr!::str_c()` `stringr!::str_conv()` `stringr!::str_count()` `stringr!::str_detect()` `stringr!::str_dup()` `stringr!::str_ends()` `stringr!::str_extract()` `stringr!::str_extract_all()` `stringr!::str_flatten()` `stringr!::str_glue()` `stringr!::str_glue_data()` `stringr!::str_interp()` `stringr!::str_length()` `stringr!::str_locate()` `stringr!::str_locate_all()` `stringr!::str_match()` `stringr!::str_match_all()` `stringr!::str_order()` A grammar for (re)shaping datas
97. Import Tidy Transform Visualise Model Communicate Visualisation Data Science: A workflow ~REALITY ~
98. Import Tidy Transform Visualise Model Communicate Visualisation Data Science: A workflow ~REALITY ~
99. An introduction to making charts with R
100. AN EXAMPLE WITH DPLYR A R package giving you a Toolbox for common data manipulation
101. AN EXAMPLE WITH GGPLOT2 A R package giving you a system for declaratively creating graphics
102. A grammar for painting data
103. Start with the data NAME ANIMAL SIZE AGE REX DOG 80 9 JERRY CAT 40 12 WOODY BIRD 20 2 PLUTO DOG 70 7
104. NAME ANIMAL SIZE AGE REX DOG 80 9 JERRY CAT 40 12 WOODY BIRD 20 2 PLUTO DOG 70 7 Start with the data # A tibble: 4 x 4 NAME ANIMAL SIZE AGE <chr> <chr> <dbl> <dbl> 1 REX Dog 80 9 2 JERRY Cat 40 12 3 WOODY Bird 20 2 4 PLUTO Dog 70 7
105. Start with the data # A tibble: 4 x 4 NAME ANIMAL SIZE AGE <chr> <chr> <dbl> <dbl> 1 REX Dog 80 9 2 JERRY Cat 40 12 3 WOODY Bird 20 2 4 PLUTO Dog 70 7 pet_data NAME ANIMAL SIZE AGE REX DOG 80 9 JERRY CAT 40 12 WOODY BIRD 20 2 PLUTO DOG 70 7
106. ggplot( data = pet_data )
107. ggplot( data = pet_data, mappings = aes( x = SIZE ) )
108. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) )
109. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_point()
110. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_point( size = 8 )
111. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_col()
112. ggplot( data = pet_data, mappings = aes( x = NAME, y = AGE ) ) + geom_col()
113. ggplot( data = pet_data, mappings = aes( x = NAME, y = AGE ) ) + geom_col() Don’t “do” the plot. Describe the plot.
114. What’s a data-driven visualisation? data-driven data-driven data-driven data-driven
115. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_point( size = 8 )
116. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_point( size = 8, color = “red” )
117. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE ) ) + geom_point( size = 8, color = c(“red”, “blue”, “green”, “yellow”) )
118. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE, color = ANIMAL ) ) + geom_point(size = 8)
119. ggplot( data = pet_data, mappings = aes( x = SIZE, y = AGE, color = ANIMAL, shape = ANIMAL ) ) + geom_point(size = 8)
120. What’s a data-driven visualisation? data-driven data-driven data-driven data-driven X position Y position Shape Outline Color Fill Color Transparency Size
121. What’s a data-driven visualisation? data-driven data-driven data-driven data-driven X position Y position Shape Outline Color Fill Color Transparency Size
122. Endless possibilities Full reproducibility
123. Exercise Menu Try to change some pieces of code Get a feel for what a chart declaration look like in ggplot2 Understand that looks is completely up to you Go to: exts.shinyapps.io/intro