Working with datasets

Stack columns with melt

stacked <- reshape2::melt(data, 
                          id.vars, 
                          measure.vars, 
                          variable.name = "variable", 
                          na.rm = FALSE, 
                          value.name = "value", 
                          factorsAsStrings = TRUE)

Note: equivalent to gather() in tidyr.

Add subtitles or units to column headers

Hmisc::label(dat, column) <- 'Some subtitle'
Hmisc::units(dat, column) <- 'mL/h'

Sort by list of columns

data[do.call('order', as.list(Data[, c('ID', 'TIME', 'MDV')])), ]

Note: equivalent to arrange() in dplyr.

Vectors

# Make sure to return a vector and not contain any lists
c(A, B, recursive = TRUE)

# Keep the dimensions of the subsetted object 
data[data[, 1] > 4, drop = TRUE]

Create unique factor based on the combination of 2 columns.

interaction(c('yellow', 'Blue'), 1:2)

## [1] yellow.1 Blue.2  
## Levels: Blue.1 yellow.1 Blue.2 yellow.2

Remove duplicated elements

# Based on one column
data[!duplicated(data[, 'ID']), ]

# Based on multiple columns
data[!duplicated(data[, c('ID', 'TIME')]), ]

Note: equivalent to distinct() in dplyr.

Compare elements

# Given
A <- 1:10
B <- 6:13

# Are A and B identical
identical(A, B)

## [1] FALSE

# Combine elements of A and B
union(A, B)

##  [1]  1  2  3  4  5  6  7  8  9 10 11 12 13

# Elements of A not in B
setdiff(A, B)

## [1] 1 2 3 4 5

# Comon elements in A and B
intersect(A, B)

## [1]  6  7  8  9 10

# ???
setequal(A,B)

## [1] FALSE

Modulus

# Modulus of 230 by 12
230 %% 12

## [1] 2

# Generate on/off infusion dosing for multiple dosing
Mod   <- Time %% Cycle            # Modulus of time by treatment cycle
Dose  <- ifelse(Mod < 14, 0, 50)  # Generate on off dosing switch

# Select rows with odd number
data[rownames(data) %% 2 == 1, ]

Create a dataset skeleton

Creates all combination of ID and TIME and adds MDV at the end.

MDV = 0
expand.grid(ID = 1:4, TIME = 0:10, include = MDV)

Match

# Common use for match()
A %in% B

# Replace COL in A based on the value of B using matching ID
A$var <- B$var[match(A$ID, B$ID)]

Note: see package fmatch for faster equivalent.

for loops

# Crashes if x not defined
for (i in 1:length(x)) { print(x[i]) }

# Can handle missing x
for (i in seq_along(x)) { print(x[i]) }

Replicate rows

# Replicate each rows 2 times
data[rep(row.names(data), times = 2), ]

# Replicate each row based on the values in a column
data[rep(row.names(data), times = data$n_rep), ]

Remove NULL elements from a list

x <-  x[-(which(sapply(x, is.null), arr.ind = TRUE))]

Aggregate

# Using formula
aggregate(DV ~ TIME, data = data, FUN = mean)

# Using by
aggregate(x = data[, 'DV'], by = list(data[, 'TIME']), FUN = mean)

Note: see also summarise() in dplyr.

Read in big data

See packages readr (read_table(), read_csv()) and data.table (fread())

Get the mean between points

Get mid point with previous row in a vectorized manner.

x[-length(x)] + (diff(x) / 2)