Demonstrate dplyr::group_by() error when a variable has class 'AsIs'

2014-05-27_dplyr-does-not-like-AsIs.r

library(dplyr)

## accept default behavior, i.e. stringsAsFactors = TRUE
## downside is that 'name' really shouldn't be a factor
df <- data.frame(name = c("jenny", "jim", "reed", "lorrie"),
                 gender = c("female", "male", "male", "female"))
df %>% group_by(gender) %>% summarize(count = n()) # this works

## protect a single character variable with I()
df <- data.frame(name = I(c("jenny", "jim", "reed", "lorrie")),
                 gender = c("female", "male", "male", "female"))
df %>% group_by(gender) %>% summarize(count = n()) # does not work!
# Error: column 'name' has unsupported type

## global protection
## downside is that 'gender' really should be a factor
df <- data.frame(name = c("jenny", "jim", "reed", "lorrie"),
                 gender = c("female", "male", "male", "female"),
                 stringsAsFactors = FALSE)
df %>% group_by(gender) %>% summarize(count = n()) # this works

## retrospective character protection
## downside is that this is annoying
df <- data.frame(name = c("jenny", "jim", "reed", "lorrie"),
                 gender = c("female", "male", "male", "female"))
df$name <- as.character(df$name)
df %>% group_by(gender) %>% summarize(count = n()) # this works