Functions

guardianapi contains functions to search and retrieve articles, tags and editions from the Guardian open data platform.

Let’s look at a few reviewers. For example, I noticed that comedy critic Brian Logan seemed to give out very few five star or one star reviews, so I wanted to see if that was true. I’ve included all his reviews from 2002–2018

library(guardianapi)
library(dplyr)
library(lubridate)
library(ggplot2)

logan_search <- gu_items(query = "profile/brianlogan")

logan_search$star_rating <- as.numeric(logan_search$star_rating)

logan_reviews <- logan_search %>% 
  filter(!is.na(star_rating), 
         web_publication_date >= as.Date("2002-01-01"),
         web_publication_date <= as.Date("2018-12-31"))

logan_reviews$year <- as.factor(year(logan_reviews$web_publication_date))

logan_summary <- logan_reviews %>%
  group_by(year, star_rating) %>%
  summarise(count = n()) %>%
  mutate(perc = count/sum(count)) %>%
  ungroup() %>%
  mutate(star_rating = factor(star_rating, levels = c(5,4,3,2,1)))

p_logan <- ggplot(data = logan_summary,
                  aes(x = year, y = count, group = star_rating)) + 
  geom_line(aes(colour = star_rating), size = 1, alpha = 0.9) + 
  scale_colour_viridis_d(name = "Rating") + 
  labs(x="Year", y="Number of Review with Rating") + 
  theme(axis.text.x = element_text(angle = 45, vjust=0.5))

p_logan

p_logan_area <- ggplot(data = logan_summary,
                  aes(x = year, y = perc, group = star_rating)) + 
  geom_area(aes(fill = star_rating)) + 
  scale_y_continuous(labels = scales::percent) + 
  scale_fill_viridis_d(name = "Rating") + 
  labs(x="Year", y="Number of Review with Rating") + 
  theme(axis.text.x = element_text(angle = 45, vjust=0.5)) 


p_logan_area

As you can see here, Brian Logan is pretty stingy with five star reviews, and didn’t give out a single five star rating in all of 2017. Likewise, he hasn’t completed panned any act with a single star since 2014.

Now let’s take a look at film critic Peter Bradshaw. I’ve used the same time span, and I’ve removed the single 0-star rating given to the 2008 film Boat Trip. There are more than four times as many film reviews from Peter Bradshaw as there are comedy reviews from Brian Logan over the same time period.

library(dplyr)
library(lubridate)
library(ggplot2)

bradshaw_search <- gu_items(query = "profile/peterbradshaw")

bradshaw_search$star_rating <- as.numeric(bradshaw_search$star_rating)

bradshaw_reviews <- bradshaw_search %>% 
  filter(!is.na(star_rating), star_rating != 0,
         web_publication_date >= as.Date("2002-01-01"),
         web_publication_date <= as.Date("2018-12-31"))

bradshaw_reviews$year <- as.factor(year(bradshaw_reviews$web_publication_date))

bradshaw_summary <- bradshaw_reviews %>%
  group_by(year, star_rating) %>%
  summarise(count = n()) %>%
  mutate(perc = count/sum(count)) %>%
  ungroup() %>%
  mutate(star_rating = factor(star_rating, levels = c(5,4,3,2,1)))

p_bradshaw <- ggplot(data = bradshaw_summary,
                  aes(x = year, y = count, group = star_rating)) + 
  geom_line(aes(colour = star_rating), size = 1, alpha = 0.9) + 
  scale_colour_viridis_d(name = "Rating") + 
  labs(x="Year", y="Number of Review with Rating") + 
  theme(axis.text.x = element_text(angle = 45, vjust=0.5))

p_bradshaw

p_bradshaw_area <- ggplot(data = bradshaw_summary,
                  aes(x = year, y = perc, group = star_rating)) + 
  geom_area(aes(fill = star_rating)) + 
  scale_y_continuous(labels = scales::percent) + 
  scale_fill_viridis_d(name = "Rating") + 
  labs(x="Year", y="Number of Review with Rating") + 
  theme(axis.text.x = element_text(angle = 45, vjust=0.5))

p_bradshaw_area

We can compare the distributions of ratings given by the two critics.


bradshaw_reviews$byline <- "Peter Bradshaw"

logan_reviews$byline <- "Brian Logan"

comp_df <- bind_rows(logan_reviews, bradshaw_reviews) %>%
  mutate(star_rating = as.numeric(star_rating))

comp_df2 <- comp_df %>%
  group_by(star_rating, byline) %>%
  summarise(count = n()) %>% group_by(byline) %>%
  mutate(perc = count/sum(count))

comp_p <- ggplot(comp_df, 
                 aes(x = star_rating, y = ..density.., fill = byline)) + 
  geom_histogram(position="dodge", bins = 5, alpha = 0.5) +
  scale_y_continuous(labels = scales::percent) +
  scale_fill_viridis_d(end = 0.9, option = "inferno") + 
  labs(x = "Star Rating", y = "", fill = "") + 
  theme(legend.position = "bottom") +
  geom_line(aes(x = star_rating, y = perc,
                colour = byline, group = byline), data = comp_df2,
            size = 1) + 
  scale_colour_viridis_d(end = 0.9, option = "inferno")  +
  guides(colour = FALSE)

comp_p

We can also use gu_content() for more general queries. For example, here’s all the articles returned for “relationships” between the two given dates:

relations <- gu_content(query = "relationships", from_date = "2018-11-30",
                        to_date = "2018-12-30")

tibble::glimpse(relations)
#> # A tibble: 170 x 44
#>    id      type  section_id section_name web_publication_da… web_title  web_url 
#>    <chr>   <chr> <chr>      <chr>        <dttm>              <chr>      <chr>   
#>  1 music/… arti… music      Music        2018-12-02 09:10:28 The 1975:… https:/…
#>  2 film/2… arti… film       Film         2018-12-18 09:48:12 Woman cla… https:/…
#>  3 lifean… arti… lifeandst… Life and st… 2018-12-10 09:00:39 I’m in a … https:/…
#>  4 politi… arti… politics   Politics     2018-12-25 18:00:00 'Special … https:/…
#>  5 footba… arti… football   Football     2018-12-19 11:08:41 Kelly Cat… https:/…
#>  6 film/2… arti… film       Film         2018-12-14 15:06:11 Sondra Lo… https:/…
#>  7 footba… arti… football   Football     2018-12-05 02:16:40 Mancheste… https:/…
#>  8 commen… arti… commentis… Opinion      2018-12-24 11:41:14 How to ge… https:/…
#>  9 books/… arti… books      Books        2018-12-15 15:00:13 Dolly Ald… https:/…
#> 10 footba… arti… football   Football     2018-12-09 18:00:20 Phoenix s… https:/…
#> # … with 160 more rows, and 37 more variables: api_url <chr>, tags <list>,
#> #   is_hosted <lgl>, pillar_id <chr>, pillar_name <chr>, headline <chr>,
#> #   standfirst <chr>, trail_text <chr>, byline <chr>, main <chr>, body <chr>,
#> #   newspaper_page_number <chr>, star_rating <chr>, wordcount <chr>,
#> #   comment_close_date <dttm>, commentable <chr>,
#> #   first_publication_date <dttm>, is_inappropriate_for_sponsorship <chr>,
#> #   is_premoderated <chr>, last_modified <chr>, newspaper_edition_date <date>,
#> #   production_office <chr>, publication <chr>, short_url <chr>,
#> #   should_hide_adverts <chr>, show_in_related_content <chr>, thumbnail <chr>,
#> #   legally_sensitive <chr>, lang <chr>, body_text <chr>, char_count <chr>,
#> #   should_hide_reader_revenue <chr>, show_affiliate_links <chr>,
#> #   sensitive <chr>, display_hint <chr>, byline_html <chr>,
#> #   live_blogging_now <chr>

Use the tag parameter to limit articles to particular sections:

relations_sex <- gu_content(query = "relationships", from_date = "2018-11-30",
                            to_date = "2018-12-30", tag = "lifeandstyle/sex")

relations_sex
#> Rows: 5
#> Columns: 40
#> $ id                               <chr> "lifeandstyle/2018/dec/10/im-in-a-rel…
#> $ type                             <chr> "article", "article", "article", "art…
#> $ section_id                       <chr> "lifeandstyle", "lifeandstyle", "life…
#> $ section_name                     <chr> "Life and style", "Life and style", "…
#> $ web_publication_date             <dttm> 2018-12-10 09:00:39, 2018-12-21 11:0…
#> $ web_title                        <chr> "I’m in a relationship with another m…
#> $ web_url                          <chr> "https://www.theguardian.com/lifeands…
#> $ api_url                          <chr> "https://content.guardianapis.com/lif…
#> $ tags                             <list> [<data.frame[10 x 13]>], [<data.fram…
#> $ is_hosted                        <lgl> FALSE, FALSE, FALSE, FALSE, FALSE
#> $ pillar_id                        <chr> "pillar/lifestyle", "pillar/lifestyle…
#> $ pillar_name                      <chr> "Lifestyle", "Lifestyle", "Lifestyle"…
#> $ headline                         <chr> "I’m in a relationship with another m…
#> $ standfirst                       <chr> "We kiss and cuddle, but he won’t go …
#> $ trail_text                       <chr> "We kiss and cuddle, but he won’t go …
#> $ byline                           <chr> "Pamela Stephenson Connolly", "Anonym…
#> $ main                             <chr> "<figure class=\"element element-imag…
#> $ body                             <chr> "<p><strong>Until last year, I identi…
#> $ newspaper_page_number            <chr> "7", "66", "83", NA, "44"
#> $ wordcount                        <chr> "387", "307", "2189", "759", "1559"
#> $ comment_close_date               <dttm> 2018-12-13 09:00:39, 2018-12-23 21:00…
#> $ commentable                      <chr> "true", "true", "true", "false", "fal…
#> $ first_publication_date           <dttm> 2018-12-10 09:00:39, 2018-12-21 11:00…
#> $ is_inappropriate_for_sponsorship <chr> "false", "false", "false", "false", …
#> $ is_premoderated                  <chr> "true", "true", "true", "false", "fal…
#> $ last_modified                    <chr> "2018-12-10T08:00:39Z", "2018-12-23T…
#> $ newspaper_edition_date           <date> 2018-12-10, 2018-12-22, 2018-12-08, N…
#> $ production_office                <chr> "UK", "UK", "UK", "UK", "UK"
#> $ publication                      <chr> "The Guardian", "The Guardian", "The …
#> $ short_url                        <chr> "https://gu.com/p/a5fad", "https://gu…
#> $ should_hide_adverts              <chr> "false", "false", "false", "false", …
#> $ show_in_related_content          <chr> "true", "true", "true", "true", "true"
#> $ thumbnail                        <chr> "https://media.guim.co.uk/35fd3d18854…
#> $ legally_sensitive                <chr> "false", "false", "false", "false", "…
#> $ sensitive                        <chr> "true", NA, NA, NA, "true"
#> $ lang                             <chr> "en", "en", "en", "en", "en"
#> $ body_text                        <chr> "Until last year, I identified as a s…
#> $ char_count                       <chr> "2192", "1729", "12190", "4404", "866…
#> $ should_hide_reader_revenue       <chr> "false", "false", "false", "false", "…
#> $ show_affiliate_links             <chr> "false", "false", "false", "false", "…