R Visualisation (base graphics)

Context: FIT1043_MOC · base-R plotting · same chart-per-type logic as matplotlib · on data frames · lab: 30_Projects/FIT1043_Labs/Week8-R-Solution.pdf Problem it solves: draw the chart matching a variable’s type, and read outliers off a boxplot.

Quick Revision

• 🎯 Trigger: visualise an R column ➔ barplot / hist / boxplot / plot chosen by data type.
• ⚡ Key Constraint: categorical → barplot (often via table()), continuous → hist/boxplot; two continuous → plot (scatter).

🔧 Minimal Working Example

H <- c(25,12,43,7,51); M <- c("Delhi","Beijing","Washington","Tokyo","Moscow")
barplot(H, names.arg=M, xlab="City", ylab="Happiness", col="blue", main="Happiness Index")

Expected output: a labelled bar chart, one bar per city.

• Bar (categorical) ➔ barplot(H, names.arg=, xlab=, ylab=, main=, col=).
• Stacked / grouped ➔ counts <- table(mtcars$vs, mtcars$gear); barplot(counts, ...); add beside=TRUE for grouped (else stacked). table() makes a frequency cross-tab.
• Histogram (continuous) ➔ hist(mtcars$hp, xlim=c(0,400), col="blue").
• Boxplot (continuous, by group) ➔ boxplot(mpg ~ cyl, data=mtcars) — formula y ~ group.
• Scatter (two continuous) ➔ plot(x=input$wt, y=input$mpg, xlab=, ylab=).

🔀 Variations

• Outliers from a boxplot ➔ outliers <- boxplot(mydata)$out returns the flagged points (IQR rule; see Measures of Spread and Boxplots).
• Plot to a file ➔ png("chart.png") → draw → dev.off() to close the device (repeat dev.off() until “null device”).

✍️ Practice

Practice 1: From mydata <- c(1,5,6,6,6,6,7,10), extract the outlier values using a boxplot.

Reference solution

mydata <- c(1,5,6,6,6,6,7,10)
outliers <- boxplot(mydata)$out
outliers   # the value(s) beyond 1.5*IQR

• Key move: boxplot(...)$out pulls the outliers the IQR rule flagged.

⚠️ Common Mistakes

• 💡 Match the chart to the type ➔ barplot is for categorical counts; use hist for a continuous distribution.
• 💡 png() needs dev.off() ➔ after plotting to a file the output is redirected; call dev.off() to finish the file and restore on-screen plotting.