Introduction to `guardianapi`

In guardianapi: Access 'The Guardian' Newspaper Open Data API

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

options("gu.API.key" = "test")

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

knitr::include_graphics("logan-plot.png")

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

knitr::include_graphics("logan-area.png")

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

knitr::include_graphics("bradshaw-plot.png")

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

knitr::include_graphics("bradshaw-area.png")

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

knitr::include_graphics("logan-bradshaw-comp.png")

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)

relations <- readr::read_rds("relations.rds")

relations

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

relations_sex <- readr::read_rds("relations_sex.rds")

tibble::glimpse(relations_sex)

guardianapi documentation built on June 24, 2019, 1:02 a.m.