Testing comparisons between data.frame and tibble objects with all.equal()

allequal-tibble-dataframe.R

# all.equal() behaves differently when presented with objects of class
# data.frame versus those with tibble class. This is important for 
# trying to rewrite the {tidyr} episode of Software Carpentry's 
# R for Reproducible Scientific Analysis lesson to include the new
# pivot_wider() and pivot_longer() columns instead of spread() and
# gather(), which have been superseded.

# Seems to be a noted problem:
# https://github.com/tidyverse/tibble/issues/287

# Software Carpentry episode:
# https://github.com/swcarpentry/r-novice-gapminder/blob/master/_episodes_rmd/14-tidyr.Rmd
# {tidyr} 1.0.0 release:
# https://www.tidyverse.org/articles/2019/09/tidyr-1-0-0/


# Read packages -----------------------------------------------------------


library(tibble)


# Create dataframes -------------------------------------------------------


# Create dataframe
df1 <- data.frame(
  col1 = LETTERS[1:5],  # factor
  col2 = 1:5,  # integer
  col3 = 6:10  # integer
)

# Copy dataframe and change third column to numeric
df2 <- df1
df2$col3 <- as.numeric(df2$col3)

# Tibble copy of first dataframe
df3 <- as_tibble(df1)

# Tibble copy of second dataframe
df4 <- as_tibble(df2)


# See dataframes ---------------------------------------------------------


str(df1)
#'data.frame':	5 obs. of  3 variables:
# $ col1: Factor w/ 5 levels "A","B","C","D",..: 1 2 3 4 5
# $ col2: int  1 2 3 4 5
# $ col3: int  6 7 8 9 10

str(df2)
#'data.frame':	5 obs. of  3 variables:
# $ col1: Factor w/ 5 levels "A","B","C","D",..: 1 2 3 4 5
# $ col2: int  1 2 3 4 5
# $ col3: num  6 7 8 9 10

str(df3)
#Classes ‘tbl_df’, ‘tbl’ and 'data.frame':	5 obs. of  3 variables:
# $ col1: Factor w/ 5 levels "A","B","C","D",..: 1 2 3 4 5
# $ col2: int  1 2 3 4 5
# $ col3: int  6 7 8 9 10

str(df4)
#Classes ‘tbl_df’, ‘tbl’ and 'data.frame':	5 obs. of  3 variables:
# $ col1: Factor w/ 5 levels "A","B","C","D",..: 1 2 3 4 5
# $ col2: int  1 2 3 4 5
# $ col3: num  6 7 8 9 10


# Compare dataframes ------------------------------------------------------


# col3 is int for df1, num for df2
all.equal(df1, df2)
#[1] TRUE

# df3 is a tibble of df1 
all.equal(df1, df3)
#[1] "Attributes: < Component “class”: Lengths (1, 3) differ (string compare on first 1) >"
#[2] "Attributes: < Component “class”: 1 string mismatch >" 

# df4 is a tibble of df2
all.equal(df2, df4)
#[1] "Attributes: < Component “class”: Lengths (1, 3) differ (string compare on first 1) >"
#[2] "Attributes: < Component “class”: 1 string mismatch >" 

# tibble versions of df1 and df2
all.equal(df3, df4)
#[1] "Incompatible type for column `col3`: x integer, y numeric"

# So comparing the two as dataframes caused no problem, but comparing the tibbles
# was stricter. Comparing data.frame to tibble had its own error message.