Tibble manipulation

Third assignment

Author

Mariana Montes

The goal of this assignment is to practice working with tables using tidyverse. Tidyverse is a collection of R packages for data wrangling and visualization (among other things). A great resource to learn how to use it is R for data Science.

For this assignment I ask that you create both a script to manipulate the table and then a Quarto file to print and cross-reference the table (combining what you learned in the second assignment).

1 Instructions

  1. Create a branch for the assignment, e.g. tibble. You will work here and only move your changes to main if and when you want to submit.

  2. Create an R script where you will insert the necessary code to do the following:

    1. Load the appropriate libraries (tidyverse and mclm).

    2. Read the brown corpus.

    3. Create an association scores table of the collocations of a word of your choosing.

    4. Save the file.

      1. OPTION A: Manipulate the table as done in class: turn it into a tibble with as_tibble(), modify columns, select some columns to show, filter the rows, rearrange the order. Then write it to a file with write_tsv().

      2. OPTION B: Save the association scores object to a file with write_assoc().

  3. Create a Quarto report where you will only load the {mclm} and {kableExtra} packages.

    1. Read the association scores object:1

      1. OPTION A: If you wrote it with write_tsv(), use read_tsv().

      2. OPTION B: If you wrote it with write_assoc(), you can either use read_tsv() or read_assoc() followed by as_tibble().

    2. If you hadn’t manipulated the table, this is the time to do so.

    3. Print the table with {kableExtra}, editing it as well if you so wish. Don’t forget to add a caption!

    4. Include some text cross-referencing the table and maybe commenting on the result.

Render Latex in table header

If you try to use Latex in a table, for example r"($\chi^2$)" to obtain \(\chi^2\), you might notice that it prints well in the interactive session but not in the rendered HTML document. This is (I think) a bug somewhere in the rendering of tables, but there is a workaround:

Somewhere in your Quarto document, paste the following text (as normal text, not as R code):

<script type="text/x-mathjax-config">MathJax.Hub.Config({tex2jax: {inlineMath: [["$","$"]]}})</script>
<script async src="https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>

This will activate parsing of Latex inside the HTML tables.

In PDF output instead you just have to add the option escape = FALSE to your kbl() call. Notice, however, that you should not have unescaped Latex characters in other elements of the table! (No underscores, for example).

2 Tips

2.1 Manipulating the table

  • Use mutate() to change the values of a column.

  • Use filter() to subset the rows based on values in the columns. You can also use the slice_ family of functions to subset with other criteria:

    • slice_head(n = 3) to select the first three rows; slice_tail(n = 5) to select the last five rows.

    • slice_sample(n = 10) to select ten random rows, slice_sample(prop = 0.5) to select a random 50% of the rows.

  • Use select() to subset the columns. You can also use rename() to rename columns without removing the rest.

  • Use arrange() to sort the tibble based on the values in a column.

An example is the code below, which starts with an assoc object (product of assoc_scores()) and ends with a tibble with a selection of rows and columns.

subsetted <- hot_assoc %>% 
  as_tibble() %>% 
  filter(PMI > 1, G_signed >= 5, endsWith(type, "nn")) %>% 
  select(type, freq = a, PMI, ends_with("signed"), OR) %>% 
  mutate(
    log_OR = log(OR),
    type = str_remove(type, "/nn")
  )

  • In line 2 we use as_tibble() to turn the assoc object into a tibble to manipulate with {tidyverse} functions.

  • In line 3 we use filter() to subset the rows that have PMI larger than 1, G_signed larger than or equal to 5, and a type ending with “nn”, i.e. nouns.

  • In line 4 we use select() to subset the columns type, a, PMI and OR as well as those ending in “signed” and at the same time rename a to “freq”.

  • In lines 5 through 8 we use mutate() to create a new column log_OR that contains the logarithm of the OR column, and we modify the type column to remove the “/nn” ending from its elements.

  • In line 1 we assign the whole operation, initially applied to hot_assoc, to a variable called subsetted.

Each operation of the pipe acts on the output of the operation before it.

2.2 Association scores

  • Use assoc_scores() after surf_cooc() to create an association scores object.

  • Use write_assoc(scores_object, filename) to save the object from your R script.

  • Use scores_object <- read_assoc(filename) to read the object in the Quarto file.

2.3 KableExtra

Check out the documentation for HTML or PDF output to learn about {kableExtra} features.

3 Git workflow

git status # check that you're on main, nothing to commit...
git branch tibble
git checkout tibble
# work on your .qmd file, render
git status # check everything is fine
git add .
git commit -m "practice with tibbles"
# you may also make several commits as you add a figure, a table...
git checkout main
git status # check everything is fine. New files should not be there
git merge tibble
# Now the .qmd file, the rendered file and the help files should be present
git push
# and send me a message!

Footnotes

  1. If you use read_tsv(), the show_col_types = FALSE argument will hide the printed output with the description of the column types, e.g. my_data <- read_tsv("filepath", show_col_types = FALSE).↩︎