###=================================================================### ###=================================================================### ### OMDS Workshop: Exploring Your Data w/ggplot2 in R ### ### by: Sarah Schwartz ### date: 2-18-16 ###=================================================================### ###=================================================================### ### ----------------- clear the global environment ---------------- ### # remove a list of the complete listing so you start with a blank slate rm(list = ls()) ### --------------- load packages you expect to use --------------- ### # For data wrangling library(plyr) # split, apply, combine, arraange, mutate library(dplyr) # subset & join library(tidyr) # gather & spread library(reshape2) # melt & cast # For plotting library(ggplot2) # the big guns! library(ggthemes) # more preset plotting themes library(RColorBrewer) # nice color pallettes # Misc library(MASS) # includes the birthwt dataset library(nlme) # includes the Orthodont dataset # http://www.r-bloggers.com/do-your-data-janitor-work-like-a-boss-with-dplyr/ # https://cran.r-project.org/web/packages/tidyr/vignettes/tidy-data.html ### ------------------ set working directory -----------------------### # this should be a FOLDER, not an specific file (unique for each project) fp = file.path("C:/Users/A00315273/Box Sync", "Office of Research Services", "OMDS", "OMDS Workshops", "OMDS ggplot2") # set the wokring directory, where to get and save 'stuff' setwd(fp) # ask to "get" or show you the working directory getwd() ### --------------------------------------------------------------- ### ### --------------------------------------------------------------- ### ### SEVEN BASIC CONCEPTS of ggplot2 ### --------------------------------------------------------------- ### ### --------------------------------------------------------------- ### # "ggplot" the main function where you specify: # (a) your dataset # (b) aesthetics: variable mappings # # Geom = Geometrics includes: point, bar, density, line, area, ... # Stat = Statistics include: # Coord = Coordinate Systems: cartesian, polar, ... # # Scales: how to plot - continuous, discrete, log, ... # # Layer: add one thing at a time # Facet: multiple panels # Saving help.search("geom_", package = "ggplot2") help.search("stat_", package = "ggplot2") help.search("coord", package = "ggplot2") help.search("theme", package = "ggplot2") help.search("scale", package = "ggplot2") ###--------------------------------------------------------------- ### ### ----------------- The Included MPG dataset ------------------ ### ###--------------------------------------------------------------- ### # this dataset is included in the ggplot2 package download # type "mpg" into the help-tab's search-box to get the documentation # see if the data is what you expected: str(mpg) # structure, compact display dim(mpg) # dimentions, # rows x # columns names(mpg) # names of the variables head(mpg) # the first few rows (subjects or cases) ### --------- Always need to declare the data & mapping -----------### # This just gives you a blank layer p0 = ggplot(data = mpg, mapping = aes(x = cty, # aes = "aesthetics" y = hwy)) summary(p0) # gives thehe summary print(p0) # creates the plot in the 'Plots' window p0 # does the same thing ### ------------------------ GEOM LAYER ---------------------------### # This adds the points toothe blank layer (blank grid) p1 = p0 + geom_point() summary(p1) p1 # You can specify the COLOR and size of ALL the points... p0 + geom_point(color= "red", size = 3) # ...or you can use a VARIABLE to color the points p2 = ggplot(data = mpg, mapping = aes(x = cty, y = hwy, color = factor(year))) + geom_point() summary(p2) p2 ### ------------------------ STAT LAYER ---------------------------### # You can add smoothed lines: # "loess" = local partial polynomial regression p1 + stat_smooth() # smooths all the points p2 + stat_smooth() # smooths each group p3 = p0 + geom_point(aes(color = factor(year))) + # color the points ONLY stat_smooth() # leave the smoother on ALL summary(p3) p3 ### ----------- Equivalent Ways to Create the Same Plot ---------- ### ### TWO steps p4 = ggplot(data = mpg, mapping = aes(x = cty, y = hwy)) p4 + geom_point(aes( color = factor(year))) + stat_smooth() ### ONE step p5 = ggplot(mpg) + geom_point(aes(x = cty, y = hwy, color = factor(year))) + stat_smooth(aes(x = cty, y = hwy)) summary(p5) p5 ### ------------------------- SCALE LAYER ------------------------ ### # Specify your OWN COLOR choices ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year))) + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) # color names # Use another VARIABLE to specify the SIZE of the ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ)) + # add size stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) ### -------------- Dealing with Over-Plotting -------------------- ### # make the points TRANSPARENT to some degree ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.5) + # 0 = clear vs. 1 = solid stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) # move the points slightly aroung with JITTER ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.4, position = "jitter") + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) # adjust the SIZE RANGE of the points ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.4, position = "jitter") + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) + scale_size_continuous(range = c(4, 10)) ### ------------------------- COORD LAYER ------------------------ ### # FLIP the x and y axses ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.4, position = "jitter") + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) + scale_size_continuous(range = c(4, 10)) + coord_flip() # instead of rectangular, you could do a POLAR coordinate system...LOL ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.4, position = "jitter") + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) + scale_size_continuous(range = c(4, 10)) + coord_polar() # more likely you many want to RESRICT the axis to ZOOOM ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = factor(year), size = displ), alpha = 0.4, position = "jitter") + stat_smooth(aes(x = cty, y = hwy)) + scale_color_manual(values = c("steelblue", "red4")) + scale_size_continuous(range = c(4, 10)) + coord_cartesian(xlim = c(15, 25), ylim = c(15, 40)) ### ------------------------- FACET LAYER ------------------------ ### # make a seperate PANEL based on a VARIABLE... ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = class, # changed the color variable size = displ), alpha = 0.4, position = "jitter") + scale_size_continuous(range = c(4, 10)) + facet_grid(year ~ .) # ROWS by year # ...or by TWO variiables ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = class, size = displ), alpha = 0.4, position = "jitter") + scale_size_continuous(range = c(4, 10)) + facet_grid(. ~ year) # COLUMNS by year ### --------------------- Add Labels & Titles -------------------- ### ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = class, size = displ), alpha = 0.4, position = "jitter") + scale_size_continuous(range = c(2, 8)) + facet_grid(drv ~ year) + labs(x = "Driving Distance per Gallon on City Roads", # x-axis y = "Driving Distance per Gallon on Highways", # y-axis size = "Displacement", # legend for size color = "Model") # legend for color ### --------------------- Changing the Theme -------------------- ### # store a plot to play with p6 = ggplot(mpg, aes(x = cty, y = hwy)) + geom_point(aes(color = class, size = displ), alpha = 0.4, position = "jitter") + scale_size_continuous(range = c(2, 8)) + facet_grid(. ~ year) + labs(x = "Driving Distance per Gallon on City Roads", # x-axis y = "Driving Distance per Gallon on Highways", # y-axis size = "Displacement", # legend for size color = "Model") # play around with default sets p6 + theme_bw() + scale_color_economist() # I use this one most of the time p6 + theme_classic() p6 + theme_dark() p6 + theme_light() p6 + theme_linedraw() p6 + theme_void() p6 + theme_minimal() # even more in another package p6 + theme_economist() # The Economist magazine p6 + theme_economist_white() p6 + theme_excel() # NEVER use this one! p6 + theme_few() p6 + theme_gdocs() # Google Docs p6 + theme_solarized() p6 + theme_stata() p6 + theme_tufte() # THE man in graphics p6 + theme_wsj() # The Wall STreet Journal # there are also preset color scales you can explore p6 + theme_economist() + scale_color_economist() ###--------------------------------------------------------------- ### ### ---------------- The Included DIAMOND dataset ---------------- ### ###--------------------------------------------------------------- ### # Take a look at this dataset (from ggplot2) str(diamonds) # structure, compact display dim(diamonds) # dimentions, # rows x # columns names(diamonds) # names of the variables head(diamonds) # the first few rows (subjects or cases) # Declare the basics p7 = ggplot(diamonds, aes(x = carat, y = price)) # All-defaut scatter plot p7 + geom_point() # change the size, transparency, and theme p7 + geom_point(size = 0.1, alpha = 0.1) + theme_bw() ### ------------------------ STAT LAYER ---------------------------### # instead of plotting EVERY point, make quare BINS of the values p7 + stat_bin2d(bins = 70) p7 + stat_density2d(aes(fill = ..level..), geom = "polygon") + # more flexible shapes coord_cartesian(xlim = c(0, 1.5), ylim = c(-50, 6000)) # zoom out a bit ###--------------------------------------------------------------- ### ### --------------- The Included FAITHFUL dataset ---------------- ### ###--------------------------------------------------------------- ### # Take a look at this dataset (from base R) str(faithful) # structure, compact display dim(faithful) # dimentions, # rows x # columns names(faithful) # names of the variables head(faithful) # the first few rows (subjects or cases) ### -------------------------- HISTOGRAM ---------------------------### # set the dataset h = ggplot(faithful, aes(x = waiting)) # all defaults... h + geom_histogram() # change the BIN width h + geom_histogram(binwidth = 30, color = "black") # play a bit more h + geom_histogram(binwidth = 10, fill = "steelblue", color = "black") ###--------------------------------------------------------------- ### ### ---------------- The Included BIRTHWT dataset ---------------- ### ###--------------------------------------------------------------- ### # Take a look at this dataset (from MASS) str(birthwt) # structure, compact display dim(birthwt) # dimentions, # rows x # columns names(birthwt) # names of the variables head(birthwt) # the first few rows (subjects or cases) ### -------------------------- HISTOGRAM ---------------------------### # Hisograms are good for one, continuous variable ggplot(birthwt, aes(x = bwt)) + geom_histogram(binwidth = 500, fill = "steelblue", color = "black") ggplot(birthwt, aes(x = bwt)) + geom_histogram(binwidth = 500, fill = "steelblue", color = "black") + facet_grid(. ~ race) # panels side-by-side ggplot(birthwt, aes(x = bwt)) + geom_histogram(binwidth = 500, fill = "steelblue", color = "black") + facet_grid(race ~ .) # panels stacked vertically ### -------------------------- BOX PLOT ----------------------------### # these are my favorite for ANOVA type situations # x is the grouping variable on which you want to compare... # y is the continuous variable you are comparing ggplot(birthwt, aes(x = factor(race), y = bwt)) + geom_boxplot() # have different boxes for every combination of smoking & race ggplot(birthwt, aes(x = interaction(factor(race), factor(smoke)), y = bwt)) + geom_boxplot() # color by one of the variables ggplot(birthwt, aes(x = factor(smoke), y = bwt, fill = factor(race))) + geom_boxplot() # Could do the reverse ggplot(birthwt, aes(x = factor(race), y = bwt, fill = factor(smoke))) + geom_boxplot() + theme_bw() ###--------------------------------------------------------------- ### ### ---------------- The provided CLIMATE dataset ---------------- ### ###--------------------------------------------------------------- ### climate = read.csv("climate.csv", header = TRUE) # Take a look at this dataset str(climate) # structure, compact display dim(climate) # dimentions, # rows x # columns names(climate) # names of the variables head(climate) # the first few rows (subjects or cases) ### -------------------------- LINE PLOT ----------------------------### # leave it to the defaults ggplot(climate, aes(Year, Anomaly10y)) + geom_line() # Draw dashed lines for confidence regions ggplot(climate, aes(Year, Anomaly10y)) + geom_line(size = 0.7, color = "black") + geom_line(aes(Year, Anomaly10y + Unc10y), linetype = "dashed", size = 0.7, color = "red") + geom_line(aes(Year, Anomaly10y - Unc10y), linetype = "dashed", size = 0.7, color = "red") + theme_gray() # add a ribbon instead ggplot(climate, aes(Year, Anomaly10y)) + geom_ribbon(aes(ymin = Anomaly10y - Unc10y, ymax = Anomaly10y + Unc10y), fill = "blue", alpha = .1) + geom_line(color = "steelblue") ###--------------------------------------------------------------- ### ### ------------------ The Included IRIS dataset ----------------- ### ###--------------------------------------------------------------- ### # Take a look at this dataset (from base R) str(iris) # structure, compact display dim(iris) # dimentions, # rows x # columns names(iris) # names of the variables head(iris) # the first few rows (subjects or cases) ggplot(data = iris, aes(x = Sepal.Length, y = Sepal.Width)) myplot = ggplot(data = iris, aes(x = Sepal.Length, y = Sepal.Width)) myplot + geom_point() ### -------------------------- BAR CHART ----------------------------### # defaults ggplot(iris, aes(x = Species, y = Sepal.Length)) + geom_bar(stat = "identity") ### ---------------------- STACKED BAR CHART ------------------------### # first we have to change from wide to long format df = melt(iris, id.vars = "Species") head(df) # then we can do it ggplot(df, aes(x = Species, y = value, fill = variable)) + geom_bar(stat = "identity") ### ---------------------- SIDE-by-SIDE BAR CHART ------------------------### ggplot(df, aes(x = Species, y = value, fill = variable)) + geom_bar(stat = "identity", position = "dodge") # this forces the bar side-by-side # RColorBrewer package has lots to choose from display.brewer.all() ggplot(df, aes(x = Species, y = value, fill = variable)) + geom_bar(stat = "identity", position = "dodge") + scale_fill_brewer(palette = "Set1") # there are lots of nice combos # compare to a box plot... ggplot(df, aes(x = variable, y = value, fill = Species)) + geom_boxplot() # Save a subset of only 1000 of the rows in the diamond dataset d = diamonds[sample(1:nrow(diamonds), 1000), ] head(d) ggplot(d, aes(x = carat, y = price, color = color)) + geom_point() ggplot(d, aes(x = clarity, fill = cut)) + geom_bar(position = "dodge") ###--------------------------------------------------------------- ### ### ------------- The Included ORTHODONT dataset ---------------- ### ###--------------------------------------------------------------- ### # this dataset is included in the nlme package download ?Orthodont str(Orthodont) # structure, compact display dim(Orthodont) # dimentions, # rows x # columns names(Orthodont) # names of the variables head(Orthodont) # the first few rows (subjects or cases) ggplot(Orthodont, aes(x = age, y = distance, group = Subject, color = Sex)) + # Color by gener geom_line() ggplot(Orthodont, aes(x = age, y = distance, group = Subject)) + geom_line() + facet_grid(. ~ Sex) # panels by gender # http://www.ats.ucla.edu/STAT/r/faq/longitudinal.htm ### --------------------------------------------------------------- ### ### --------------------------------------------------------------- ### ### SAVING YOUR PLOT ### --------------------------------------------------------------- ### ### --------------------------------------------------------------- ### # If the plot is on your screen ggsave("filename.png") # if the plot is saved to "plot1" ggsave(plot1, file = "~/path/to/figure/filename.png") # you can set the size ggsave(file = "/filename.png", width = 6, height =4) # you can use any format (pdf, png, svg, jpg) ggsave(file = "filename.eps") ggsave(file = "filename.jpg") ggsave(file = "filename.pdf")