Working with data frames

Basic principles to work with data frame in base R, Tidyverse, and data.table.

Set up

Attach important packages. For a comprehensive comparison see this blog.

library(tidyverse)
library(data.table)

Load data. You can go here to take a look at some fake data in different formats

link_data <- "https://github.com/PovcalNet-Team/Rtraining/raw/main/data/ago_2018.csv"

df <- read.csv(link_data) # base
tb <- read_csv(link_data)  # tidyverse 
dt <- fread(link_data) # data.table

We could have done also this

df <- read.csv(link_data) # base
tb <- as.tibble(df)
dt <- as.data.table(tb)

Basic operations

Filter rows

Keep rows using indices

filter <- c(3:4)

Base R

df[filter,]

   area   welfare    weight
3 rural 213.44287 398.89557
4 rural 423.68354 538.45697

df[filter] # This does not work

Error in `[.data.frame`(df, filter): undefined columns selected

Tidyverse

tb[filter,]

# A tibble: 2 x 3
  area  welfare weight
  <chr>   <dbl>  <dbl>
1 rural    213.   399.
2 rural    424.   538.

slice(tb, filter) # same

# A tibble: 2 x 3
  area  welfare weight
  <chr>   <dbl>  <dbl>
1 rural    213.   399.
2 rural    424.   538.

data.table

dt[filter,]

    area   welfare    weight
1: rural 213.44287 398.89557
2: rural 423.68354 538.45697

# This works. In data.frame does not. 
dt[filter] # same. 

    area   welfare    weight
1: rural 213.44287 398.89557
2: rural 423.68354 538.45697

For the sake of time, I’ll incorporate all the examples later.

Keep rows using logical expressions

Base R

x <- df[df$area == "urban",]
x[1:3,] 

     area   welfare    weight
222 urban 665.10899 641.75787
223 urban 489.98341 112.15016
224 urban  75.45557 317.06293

Tidyverse

tb |> 
  filter(area == "urban") |> 
  slice(1:3)

# A tibble: 3 x 3
  area  welfare weight
  <chr>   <dbl>  <dbl>
1 urban   665.    642.
2 urban   490.    112.
3 urban    75.5   317.

data.table

# data.table way and No need of $
 
dt[area == "urban"
   ][1:3]

    area   welfare    weight
1: urban 665.10899 641.75787
2: urban 489.98341 112.15016
3: urban  75.45557 317.06293

# Tidyverse syntax works with data.table
dt |> 
  filter(area == "urban") |> 
  slice(1:3)

    area   welfare    weight
1: urban 665.10899 641.75787
2: urban 489.98341 112.15016
3: urban  75.45557 317.06293

# but data.table syntax does not with tidyverse
tb[area == "urban"][1:3]

Error in `[.tbl_df`(tb, area == "urban"): object 'area' not found