R Data Frames and IO

Context: FIT1043_MOC · R’s table = bound vectors · the wrangling/exploration workhorse · R’s answer to a pandas DataFrame Problem it solves: build/read a data frame, audit it, extract/sort/merge/aggregate rows and columns, and write it back.

Quick Revision

• 🎯 Trigger: tabular data in R ➔ data.frame(); audit with str/summary; slice with [row, col] and the $ operator.
• ⚡ Key Constraint: df[i] selects a column; a row needs the trailing comma df[i, ] — the comma placement changes everything.

🔧 Minimal Working Example

names <- c("Bill","Ted","Henry","Joan"); ages <- c(76,82,104,78)
myTable <- data.frame(names, ages)
names(myTable) <- c("Names","Ages")   # rename columns
dim(myTable)      # [1] 4 2   (nrow=4, ncol=2)
str(myTable)      # column names + data types
summary(myTable)  # per-column summary stats

Expected output: a 4×2 table; dim → 4 2; str/summary describe types and distributions.

• Build / rename ➔ data.frame(v1, v2, …); names(df) <- c(...).
• Audit ➔ nrow / ncol / dim; str(df) (types); summary(df); head(df) / tail(df); stats min/mean/sd on df$col.
• Extract ➔ column df["Ages"] or df$Ages; multiple df[c("Names","Ages")]; by index df[2]; row df[1, ]; cell df[1,2]; block df[3:4, 2:3].
• Sort ➔ df[order(df$Ages), ] (asc) / order(df$Ages, decreasing=TRUE); multi-key order(df$Ages, df$Heights).
• Combine ➔ merge(a, b, by="Names") (join on key); rbind(a, b) (stack rows — same columns).
• Aggregate ➔ aggregate(mtcars, by=list(cyl, vs), FUN=mean) (group → summary).

🔀 Variations

• I/O ➔ getwd() / setwd("D:/Folder"); write.csv(df, "F.csv"); df <- read.csv("F.csv", header=TRUE, stringsAsFactors=TRUE) — header names the columns from line 1, stringsAsFactors makes text columns categorical (always pass it; see Categorical Summaries and Cross-Tabulation in R).
• Libraries & datasets ➔ install.packages("moments") once, then library(moments); data() lists built-ins, data(mtcars) loads one; head(df, 10) overrides the default 6 rows; View(df) opens the RStudio spreadsheet viewer.

Logical referencing — filtering rows by a condition

heart <- read.csv("heart.csv", header = TRUE, stringsAsFactors = TRUE)
heart[heart$SEX == 0, "AGE"]                    # AGE column, rows where SEX is 0
heart[heart$CHOL < 150, c("SEX","CP")]          # two columns, rows meeting the condition
heart[heart$SEX == 0 & heart$CHOL < 150, ]      # AND — all columns, both conditions
heart[heart$SEX == 0 | heart$CHOL < 150, ]      # OR

• Mechanism ➔ heart$SEX == 0 evaluates to a logical vector the length of the data frame; placing it in the row slot keeps every TRUE position ➔ conditions compose with & (AND) and | (OR), element-wise.
• Row slot, column slot ➔ the condition goes before the comma; the columns go after.

Adding and deleting columns

heart$AC <- heart$AGE / heart$CHOL   # ADD: assign to a name that doesn't exist yet
heart$AC <- NULL                     # DELETE: assign the special value NULL

• Vectorised ➔ the arithmetic runs element-wise over all rows at once; no loop needed.

✍️ Practice

Practice 1: Load mtcars, show its first rows, and the mean of every numeric column grouped by cyl and vs.

Reference solution

data(mtcars)
head(mtcars)
aggregate(mtcars, by = list(mtcars$cyl, mtcars$vs), FUN = mean)

• Key move: aggregate(..., by=list(...), FUN=mean) is R’s group-by-then-summarise.

Practice 2: From heart.csv, show RESTECG and HD for rows 10–50; then add a column AC = AGE/CHOL, inspect it, and remove it again.

Reference solution

heart <- read.csv("heart.csv", header = TRUE, stringsAsFactors = TRUE)
nrow(heart); ncol(heart)          # 303 rows, 14 columns
heart[10:50, c("RESTECG","HD")]
heart$AC <- heart$AGE / heart$CHOL
head(heart)
heart$AC <- NULL

• Key move: 10:50 in the row slot is a range of positions; a condition in that same slot would be logical referencing. Deleting a column is <- NULL, not rm() (which removes whole objects from the workspace).

⚠️ Common Mistakes

• 💡 Comma = row vs column ➔ df[2] is column 2; df[2, ] is row 2; df[2,3] is one cell.
• 💡 = vs == inside a filter ➔ heart[heart$SEX = 0, ] is an error; comparison needs the double equals.
• 💡 setwd before read/write ➔ relative filenames resolve against the working directory; set it (or pass a full path) first.
• 💡 str may show Factor ➔ character columns can load as factors; check with str() before treating them as plain text.