FISH 6002: Week 9 - Data display 3

Week 9 fish 6002

Transcript

1. Week 9: Displaying Data Visually 3 FISH 6000: Science Communication for Fisheries Brett Favaro 2017 This work is licensed under a Creative Commons Attribution 4.0 International License
2. Week 9: •Colour basics •Increasing data density with colours •Increasing data density with symbols and annotation •Multipanel plots, and emphasizing contrasts • Messing with margins
3. 1. Colours, symbols, and annotation Colours can -Beautify -Organize -Increase data density
4. First, the basics… Recall: ggplot base
5. # name: colors() plot(log10(weight) ~ log10(length), data=perch,xlim=c(1.4,2.5), pch=16, col="red") # col changes the fill colour of points # https://www.rapidtables.com/web/color/RGB_Color.html
7. # name: colors() plot(log10(weight) ~ log10(length), data=perch,xlim=c(1.4,2.5), pch=16, col="red") # col changes the fill colour of points # number … col=2) # hex … col="#FF0000") # https://www.rapidtables.com/web/color/RGB_Color.html
8. plot(log10(weight) ~ log10(length), data=perch,xlim=c(1.4,2.5), pch=24, col="red", bg="blue")
9. Changing other colours par(col.axis = "darkgreen", col.lab = "purple", bg = "cyan") plot(log10(weight) ~ log10(length), data=perch,xlim=c(1.4,2.5), pch=24, col="red", bg="black") Use the par() command Every plot thereafter will have these settings… old_par <- par() # Always put this at the start of your code, if you’re messing with base plots… par(old_par) #... So you can put this line after to reset them
10. https://www.edwardtufte.com/tufte/ Data ink -------------------------- Total ink used in graphic Data ink ratio = Fill up the plot with meaningful ink – not decorative ink # of pieces of data -------------------------- Area of graphic Data density = Edward Tufte But… colours are not for decoration
11. Colours can be used to group variables Continuous Continuous 3rd axis: Discrete (here: gear type) Switch to ggplot
12. a <- ggplot(data = perch, aes(x = length, y = weight)) + geom_point(shape=1) + scale_y_log10() + scale_x_log10(limits = c(25, 300), breaks= seq(from=25, to=300, by=75)) + labs(x = "Length (mm)", y = "Weight (g)") + theme_bw() + theme(axis.text = element_text(size=16), axis.title = element_text(size=18)) print(a) Recall
13. a <- ggplot(data = perch, aes(x = length, y = weight, col=combinedgear)) + geom_point(shape=1) + scale_y_log10() + scale_x_log10(limits = c(25, 300), breaks= seq(from=25, to=300, by=75)) + labs(x = "Length (mm)", y = "Weight (g)") + theme_bw() + theme(axis.text = element_text(size=16), axis.title = element_text(size=18)) print(a) Specify your third variable as an aesthetic
14. Consider: Vs. All additions must have meaning, and be clear Here, both cases suffer from overplotting. Park that thought for a minute – we’ll talk about a solution later.
15. Fix the legend a + scale_colour_manual(name="Gear type", labels=c("Beach seine", "Crayfish trap", "Electrofishing", "Fyke net", "Minnow trap", "Trammel net", "Vertical gillnet"), values=c(1:7))
16. Colour palettes http://colorbrewer2.org Single hue Multi-hue Divergent Sequential - Negative and positive values Temperature (cold to hot) Opinion: Disagree – Agree - Value bound at zero Temperature (K) Size Qualitative - Categorical, unordered values Country Species
19. Which would make sense for gear type? Single hue Multi-hue Divergent Qualitative
20. display.brewer.pal("Set1", n=7) # Show the palette mypal <- brewer.pal("Set1", n=7) mypal "#E41A1C" "#377EB8" "#4DAF4A" "#984EA3" "#FF7F00" "#FFFF33" "#A65628"
22. This allows you to try out many colour palettes
23. http://derekogle.com/fishR/data/data-html/Pallid.html
25. Let’s plot these data Data density low Let’s add weight
26. Recall Apply a colour ramp: Heavier fish are red, lighter = green Quiz: * Is this sequential, diverging, or qualitative? * Have I used an appropriate scale here? b <- ggplot(pallid, aes(x = tl, y = sl)) + geom_point(aes(colour=w), size=4) + scale_colour_gradient(high="red", low="green") + theme_bw() + xlab("Total length (mm)") + ylab("Standard length (mm)") + theme(axis.text=element_text(size=12), axis.title = element_text(size=14) ) Normally I wouldn’t use diverging for this. I’m about to illustrate something…
28. https://www.nature.com/nmeth/journal/v8/n6/full/nmeth.1618.html
29. http://geog.uoregon.edu/datagr aphics/color_scales.htm Use dichromat package to simulate colourblindness, and select and test palettes
30. Why not red-green?
31. Why not red-green?
32. With blue- red, diverging scale remains visible
33. Good fix: blue-red. Diverging scale remains visible
34. d <- ggplot(pallid, aes(x = tl, y = sl)) + geom_point(aes(colour=w), size=8) + scale_colour_gradient(high=dichromat("darkred", type = "tritan"), low=dichromat("green", type = "tritan")) + theme_bw() + xlab("Total length (mm)") + ylab("Standard length (mm)") + theme(axis.text=element_text(size=12), axis.title = element_text(size=14), title = element_text(size=16) ) + ggtitle("Tritan - blue appears green") dichromat("red", type = "tritan") "#F35B5B" Dichromat takes a colour you input, and outputs a hex code for the colour you would see if you had this form of colourblindness
35. Colour by location * Is this sequential, diverging, or qualitative? * Have I used an appropriate scale here? This time, we learned something new. Location and fish size are related
38. What is this doing? To illustrate:
39. Most palettes are not perfect Be mindful: is the figure at all interpretable without colour? If not, may need new design!
40. Add even more data density Point size proportional to weight geom_point(aes( color=loc, size=w))
41. Let’s pack in even MORE information geom_point(aes(color=loc, size=w, shape=status)) # of pieces of data -------------------------- Area of graphic As high as possible, without becoming confusing. How are we doing here? Recall data density:
42. Let’s pack in even MORE information Adding text or shapes to plots Two types of annotation: - Manual label or shape - From data a + annotate("text", x = 900, y = 1400, label="A", size=14)
43. a + annotate("text", x = c(900,1000), y = c(1400, 1300), label=c("A", “B”), size=14) Add multiple bits of text
44. a + annotate("segment", x = 1200, xend = 1400, y = 900, yend = 1000, colour = "red", size=3, alpha=0.6) Add a line
45. a + annotate("segment", x = 1200, xend = 1400, y = 900, yend = 1000, colour = "red", size=3, alpha=0.6, arrow=arrow()) Add an arrow
46. a + annotate("rect", xmin = 1200, xmax = 1400, ymin = 900, ymax = 1000, colour = "red", size=3, alpha=0) + annotate("rect", xmin = 1200, xmax = 1400, ymin = 750, ymax = 800, colour = "blue", size=3, alpha=1, fill="black”) Add shapes!
47. Add horizontal or vertical lines a + geom_hline(yintercept=1100, color="orange", size=1) + geom_vline(xintercept=1400, color="red", size=1)
48. Annotation: Adding raster images https://cran.r-project.org/web/packages/grImport/vignettes/import.pdf
49. Add raster images
50. As-is, these add no information…
51. Adding text or shapes to plots Two types of annotation: - Manual label or shape - From data Sometimes you may want to add text to every label E.g. plot names of species on a scatterplot Let’s add the year to each dot in this plot
52. a + geom_text(label=year(pallid$date))
53. a + geom_text_repel(label=year(pallid$date))
54. Recap This plot shows: - Total length - Standard length - Year of capture - Death status - Weight - Location captured Purpose: Demonstrate how to add each of these features. This plot is probably too confusing.
57. Next: Recall… What is this problem called?
58. 2. Multipanel plots Sometimes you want to break up one plot into multiple panels Use to emphasize contrasts Use to reduce overplotting
59. https://www.nature.com/article s/s41598-017-09084-6 Facets can make graphs easier to interpret Note different scales across each group of species Troudet et al. (2017)
61. a + facet_grid(cols=vars(combinedgear)) What do you notice about the colours in the plot below?
62. a + facet_grid(cols=vars(combinedgear)) What do you notice about the colours in the plot below? Drop minnow…
64. How do these differ?
65. http://derekogle.com/fishR/data/data-html/SimonsonLyons.html Example: We want 4th 1st 2nd 3rd Number caught
66. What is wrong with this picture??
67. Still wrong: why? More than one value per species per stream
68. Tufte term: “Small multiples” Facet by stream
69. Facet by stream, logged (what happened to zeroes?)
70. Facet by stream, allowing scales to vary
71. Facet by species Which is easier to interpret? What is the goal of the plot?
72. Now, we look at population estimates
73. Emphasis: Contrasting species abundance across streams
74. Emphasis: Comparing species of each species WITHIN streams
75. Sometimes you may want to render more than one plot at a time
76. multiplot is a function A function is a piece of code that contains a series of instructions that you can call repeatedly throughout your script From: http://www.cookbook- r.com/Graphs/Multiple_graphs_on_one_page_(ggplot2)/ By running it at the top of the script, we can call it repeatedly throughout
79. http://onlinelibrary.wiley.com/doi/10.1111/j.1755-263X.2012.00294.x/full
81. http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0113118
82. https://www.economist.com/blogs/graphicdetail/2017/10/daily-chart-18
83. https://www.economist.com/blogs/graphicdetail/2017/10/daily-chart-18
84. 3. Model Output Returning to the pallid dataset, plot: y = standard length (sl) x = total length (tl)
88. CI = 50% CI = 95% CI = 99% CI = 100% More info: http://ggplot2.tidyverse.org/reference/geom_smooth.html
89. By default, ggplot2 won’t extend regression line beyond the range of data. That’s good
90. Key: ggplot2 has functions to plot common models. In base plot, you run the model separately, and plot it manually
91. Go to: http://motioninsocial.com/tufte/ Self-study • Work through entire website • Note differences between ggplot and base plot examples • Try recreating some of these plots with your own data, or with data from somewhere else
92. MINOR ASSIGNMENT 3 • Make a publication-quality plot from these data • This should be a two-variable (minimum) plot, made in either ggplot or base ❑ Select an appropriate type of plot for the data ❑Format the axes and any legends used with high-quality, descriptive labels ❑Modify at least one colour, or apply a colour gradient; and/or; ❑Use faceting or multipanel plotting ❑ have a reasonable data:ink ratio, and data density • All elements necessary for publication should be present (Panels should be counted off with A, B, C, etc)
93. SAVE this plot as a TIFF, with lzw compression, at 300-600 dpi (i.e. make it suitable for publication in Plos One) Deliverables: - One PDF, outputted from the PACE system (https://pacev2.apexcovantage.com/), containing your plot - A .txt file with a figure caption enclosed
95. • Assessment: • Did you submit a graph that meets minimal graphical parameters? (Answer is yes, if PACE outputs a PDF) /4 – Did you include a figure caption /1? • Scientific basics: /5 • Appropriate choice of plot type? • Sensible variables on X and Y? • Graphical elements /5 • Informative axes? (Good titles, large font?) • Graphical elements worked to enhance, not just decorate the plot