A simple example

The following example shows one use-case. Using Anderson’s iris data, we calculate the regression of Petal.Length on Sepal.Length for each Species and then merge this slope coefficient back into the original data.

coefs  <- lapply(split(iris, iris$Species),
                 function(dat) lm(Petal.Length~Sepal.Length, dat)$coef)
coefs <- do.call("rbind",coefs)
coefs <- as.data.frame(coefs)
coefs$Species <- rownames(coefs)
coefs
##            (Intercept) Sepal.Length    Species
## setosa       0.8030518    0.1316317     setosa
## versicolor   0.1851155    0.6864698 versicolor
## virginica    0.6104680    0.7500808  virginica
library(lookup)
iris = transform(iris,
                 slope1 = lookup(iris$Species, coefs$Species, coefs[,"Sepal.Length"]),
                 slope2 = vlookup(iris$Species, coefs, "Species", "Sepal.Length"))
head(iris)
##   Sepal.Length Sepal.Width Petal.Length Petal.Width Species    slope1    slope2
## 1          5.1         3.5          1.4         0.2  setosa 0.1316317 0.1316317
## 2          4.9         3.0          1.4         0.2  setosa 0.1316317 0.1316317
## 3          4.7         3.2          1.3         0.2  setosa 0.1316317 0.1316317
## 4          4.6         3.1          1.5         0.2  setosa 0.1316317 0.1316317
## 5          5.0         3.6          1.4         0.2  setosa 0.1316317 0.1316317
## 6          5.4         3.9          1.7         0.4  setosa 0.1316317 0.1316317

Admittedly, a better way to approach this problem would be with the dplyr package and the group_by and summarize functions. But, this example does not depend on external packages.

History

I wrote the lookup() function for my own personal use sometime before 2005. When Jenny Bryan posted vlookup() on Twitter, I modified her version and decided there would be value in making these functions widely available in a package.