Add VT2-examples: all 6 chapters of 'Visualize This' (110 files) — data scraping, ETL, trends, categorical, multivariate, spatial, distribution analysis

VT2-examples/vt2-ch4-codedata-dir/trends.R

@@ -0,0 +1,158 @@
+#
+# Bar chart
+#
+
+# Load data
+winners <- read.csv("data/hot-dog-contest-winners.csv")
+head(winners)
+
+# Default bar chart
+barplot(winners$hotdogs)
+
+# Year labels
+barplot(winners$hotdogs, names.arg=winners$year)
+
+# Axis labels
+barplot(winners$hotdogs, 
+        names.arg=winners$year,
+        border=NA,
+        space=.1,
+        xlab="Year", 
+        ylab="Hot dogs and buns (HDB) eaten",
+        main="Contest Winners")
+
+
+
+
+#
+# Line chart
+#
+
+# Graphical parameter to make axis labels horizontal
+par(las=1)
+
+# Line chart with plot()
+plot(winners$year, winners$hotdogs, 
+     type="l",
+     xlab="",
+     ylab="Hot dogs and buns (HDB) eaten",
+     main="Nathan's Hot Dog Eating Contest, Winning Results, 1980-2023")
+
+par(mfrow=c(3,2), las=1)
+for (currtype in c("p", "l", "b", "c", "o", "n")) {
+    plot(winners$year, winners$hotdogs, type=currtype,
+         xlab="", ylab="",
+         main=paste0("type='", currtype, "'"))
+}
+
+# Text
+text(2001, 50, 
+     "In 2001, Takeru Kobayashi nearly doubled the previous record with 50 HDBs.",
+     pos=2)
+
+# Text with line breaks
+text(2001, 48, 
+     "In 2001, Takeru Kobayashi 
+     nearly doubled the previous 
+     record with 50 HDBs.",
+     pos=2)
+
+# Find documentation
+?text
+
+# Wrapped text
+anno <- "In 2001, Takeru Kobayashi nearly doubled the previous record with 50 HDBs."
+text(2001, 48, 
+     paste(strwrap(anno, 30), collapse="\n"),
+     pos=2)
+
+
+#
+# Dot plot
+#
+
+par(las=1)
+plot(winners$year, winners$hotdogs, 
+     type="p",
+     bty="n",
+     xlab="",
+     ylab="Hot dogs and buns (HDB) eaten",
+     main="Nathan's Hot Dog Eating Contest, Winning Results, 1980-2023")
+
+
+
+
+#
+# Step chart
+#
+
+# Load data
+postage <- read.csv("data/us-postage.csv")
+
+# Summary
+head(postage)
+summary(postage)
+
+# Plot as regular time series
+par(las=1)
+plot(postage$Year, postage$Price, type="l",
+     xlab="Year", ylab="Postage Rate (Dollars)",
+     main="US Postage Rates for Letters, First Ounce, 1863-2023")
+
+# Default step chart
+plot(postage$Year, postage$Price, type="s")
+
+# Add labels
+par(las=1)
+plot(postage$Year, postage$Price, type="s", 
+     xlab="Year", ylab="Postage Rate (Dollars)",
+     main="US Postage Rates for Letters, First Ounce, 1863-2023")
+
+# Change limits ("zoom")
+postagesub <- postage[postage$Year >= 1991,]
+head(postagesub)
+
+par(las=1)
+plot(postagesub$Year, postagesub$Price, type="s",
+     xlab="Year", ylab="Postage Rate (Dollars)",
+     main="US Postage Rates for Letters, First Ounce, 1991-2023")
+
+
+
+
+#
+# Smoothing
+#
+
+# Load data.
+# install.packages("babynames")
+library(babynames)
+nathan <- babynames[babynames$name == "Nathan" & babynames$sex == "M", ]
+beatrice <- babynames[babynames$name == "Beatrice" & babynames$sex == "F",]
+
+# Line chart, no smoothing
+par(las=1, mar=c(4,5,3,2))
+plot(nathan$year, nathan$prop, 
+     type="l", 
+     xlab="", ylab="Proportion")
+lines(beatrice$year, beatrice$prop, 
+      lwd=2)
+
+# Splines
+natcoords <- spline(nathan$year, nathan$prop)
+beacoords <- spline(beatrice$year, beatrice$prop)
+par(las=1, mar=c(4,5,3,2))
+plot(natcoords,
+     col="#ac5c5a",
+     type="l",
+     xlab="", ylab="Proportion")
+lines(beacoords, lwd=2, col="#176572")
+
+natcoords <- spline(nathan$year, nathan$prop, n=20)
+beacoords <- spline(beatrice$year, beatrice$prop, n=20)
+par(las=1, mar=c(4,5,3,2))
+plot(natcoords,
+     col="#ac5c5a",
+     type="l",
+     xlab="", ylab="Proportion")
+lines(beacoords, lwd=2, col="#176572")