Example: Sorting

Sort values or reorder observations?

sort() returns values in order. order() returns the positions that put values in order. Use those positions to reorder complete rows of a data frame, keeping each observation’s identifier and measurements together.

Sort a vector

heights <- c(18.5, 21.0, 16.2)
sort(heights)
order(heights)

[1] 16.2 18.5 21.0
[1] 3 1 2

The smallest value is in position 3, followed by positions 1 and 2. Therefore, heights[order(heights)] gives the same ordered values as sort(heights).

Reorder complete rows

Create a small table directly in R:

measurements <- data.frame(
  ID = c("01", "02", "03", "04"),
  Year = c(2024L, 2023L, 2024L, 2023L),
  Height_m = c(18.5, 21.0, 16.2, NA),
  stringsAsFactors = FALSE
)
measurements[order(measurements$Height_m, na.last = TRUE), ]

The row order is 03, 01, 02, 04; the unavailable height comes last. The comma in [rows, columns] matters: leaving the column position empty retains every column.

To keep this result, assign it to another object:

by_height <- measurements[
  order(measurements$Height_m, na.last = TRUE),
]

Do not sort the height column and assign it back on its own. That would attach heights to different IDs and silently change the observations.

Use more than one sorting column

by_year <- measurements[
  order(measurements$Year, measurements$ID),
]
by_year

Rows are ordered by year first; IDs break ties within each year. Here the ID order is 02, 04, 01, 03. Use decreasing = TRUE to reverse the ordering. For order(), na.last = TRUE keeps missing values at the end, whereas na.last = NA removes their positions; choose explicitly.

Check the result

nrow(by_year) == nrow(measurements)
identical(sort(by_year$ID), sort(measurements$ID))

Both checks should return TRUE: this operation changes row order, not the set of observations. Real regional data repeat each ID across years, so use ID and Year when identifying an observation. Sorting is not a method for removing duplicates or repairing mismatched values.

Characters and factors

Keep identifiers as text to preserve leading zeros. Create a factor explicitly when its levels represent categories. Reordering rows does not change a factor’s level order; specify levels = ... when that order matters for plots or models.

For additional examples, see sorting in the Base R course and the R documentation for order().