Tidytuesday
Jack Baumann
11/12/2019
R Markdown
library(tidyverse)## ── Attaching packages ───────────────────────────── tidyverse 1.2.1 ──## ✔ ggplot2 3.2.1 ✔ purrr 0.3.2
## ✔ tibble 2.1.3 ✔ dplyr 0.8.3
## ✔ tidyr 0.8.3 ✔ stringr 1.4.0
## ✔ readr 1.3.1 ✔ forcats 0.4.0## ── Conflicts ──────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()theme_set(theme_light())pizza_jared <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-10-01/pizza_jared.csv")## Parsed with column specification:
## cols(
## polla_qid = col_double(),
## answer = col_character(),
## votes = col_double(),
## pollq_id = col_double(),
## question = col_character(),
## place = col_character(),
## time = col_double(),
## total_votes = col_double(),
## percent = col_double()
## )pizza_barstool <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-10-01/pizza_barstool.csv")## Parsed with column specification:
## cols(
## .default = col_double(),
## name = col_character(),
## address1 = col_character(),
## city = col_character(),
## country = col_character()
## )## See spec(...) for full column specifications.pizza_datafiniti <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-10-01/pizza_datafiniti.csv")## Parsed with column specification:
## cols(
## name = col_character(),
## address = col_character(),
## city = col_character(),
## country = col_character(),
## province = col_character(),
## latitude = col_double(),
## longitude = col_double(),
## categories = col_character(),
## price_range_min = col_double(),
## price_range_max = col_double()
## )answer_orders <- c("Never Again", "Poor", "Average", "Good", "Excellent")
by_place_answer <- pizza_jared %>%
mutate(time = as.POSIXct(time, origin = "1970-01-01"),
date = as.Date(time),
answer = fct_relevel(answer, answer_orders)) %>%
group_by(place, answer) %>%
summarize(votes = sum(votes)) %>%
mutate(total = sum(votes),
percent = votes / total,
answer_integer = as.integer(answer),
average = sum(answer_integer * percent)) %>%
ungroup()
by_place <- by_place_answer %>%
distinct(place, total, average)by_place_answer %>%
filter(as.integer(fct_reorder(place, total, .desc = TRUE)) <= 16,
answer != "Fair") %>%
mutate(place = glue::glue("{ place } ({ total })"),
place = fct_reorder(place, average)) %>%
ggplot(aes(answer, percent)) +
geom_col() +
facet_wrap(~ place) +
scale_y_continuous(labels = scales::percent) +
theme(axis.text.x = element_text(angle = 90, hjust = 1)) +
labs(x = "",
y = "% of respondents",
title = "What is the most popular pizza place in Open Stats meetup?",
subtitle = "Only the 16 pizza places with the most respondents. # respondents shown in parentheses.")
Flore’s pizza place is the most popular in open stats because it had the most respondents at 123. Th closest amount of respondents for any other pizza place was 61. It also had a small percentage of bad reviews per respondents.
library(broom)
# THIS TRICK DOESN'T WORK
# tidy(lm(c(1, 2, 3, 4, 5) ~ 1, weights = c(100, 300, 100, 200, 150)), conf.int = TRUE)
t_test_repeated <- function(x, frequency) {
tidy(t.test(rep(x, frequency)))
}
by_place_answer %>%
filter(total >= 3) %>%
group_by(place, total) %>%
summarize(t_test_result = list(t_test_repeated(answer_integer, votes))) %>%
ungroup() %>%
unnest(t_test_result) %>%
select(place, total, average = estimate, low = conf.low, high = conf.high) %>%
top_n(16, total) %>%
mutate(place = fct_reorder(place, average)) %>%
ggplot(aes(average, place)) +
geom_point(aes(size = total)) +
geom_errorbarh(aes(xmin = low, xmax = high)) +
labs(x = "Average score (1-5 Likert Scale)",
y = "",
title = "What is the most popular pizza place in Open Stats meetup?",
subtitle = "Only the 16 pizza places with the most respondents.",
size = "# of respondents")
The best rated pizza place from the graph is Tappo but also only had about 25 respondents. The most popular pizza place by amount of respondents and average rating was Flore’s at about a average of a 4 for rating with close to 125 respondents.
# Don't bother comparing them, this is a bad graph
pizza_barstool %>%
select(place = name,
barstool_total = review_stats_all_count,
barstool_average = review_stats_all_average_score) %>%
inner_join(by_place, by = "place") %>%
group_by(place) %>%
filter(n() == 1) %>%
ungroup() %>%
filter(barstool_total >= 5,
total >= 5) %>%
ggplot(aes(average, barstool_average)) +
geom_point() +
labs(x = "Meetup",
y = "Barstool")
pizza_barstool %>%
top_n(50, review_stats_all_count) %>%
ggplot(aes(price_level, review_stats_all_average_score, group = price_level)) +
geom_boxplot()
pizza_barstool %>%
filter(review_stats_all_count >= 50) %>%
mutate(name = fct_reorder(name, review_stats_all_average_score)) %>%
ggplot(aes(review_stats_all_average_score, name, size = review_stats_all_count)) +
geom_point() +
labs(x = "Average rating",
y = "",
size = "# of reviews",
title = "Barstool Sports ratings of pizza places",
subtitle = "Only places with at least 50 reviews")
pizza_barstool %>%
filter(review_stats_all_count >= 20) %>%
mutate(city = fct_lump(city, 3)) %>%
add_count(city) %>%
mutate(city = glue::glue("{ city } ({ n })")) %>%
ggplot(aes(city, review_stats_all_average_score)) +
geom_boxplot() +
labs(title = "Do pizza ratings differ across cities?",
subtitle = "Only pizza places with at least 20 reviews")
Yes pizza ratings differ across cities. As seen on the graph Boston and New York seem to stay relatively similiar in rating comparison but, Brookelyn seems to have a better average of ratings out of atleast 20 reviews. The other cities seem to have a worse average of ratings compared to these main cities.
pizza_cleaned <- pizza_barstool %>%
select(place = name,
price_level,
contains("review")) %>%
rename_all(~ str_remove(., "review_stats_")) %>%
select(-contains("provider"))
pizza_cleaned %>%
filter(critic_count > 0) %>%
ggplot(aes(critic_average_score, dave_average_score)) +
geom_point() +
geom_abline(color = "red") +
geom_smooth(method = "lm") +
labs(title = "Does Barstool Sports' Dave agree with the critics?",
x = "Critic average score",
y = "Dave score")
The graph shows that Barstool Dave does not agree with the critics. It seems that Dave is a litle more tough when rating than the critics. The critics will have a high rating and most of the time dave will be more tough and have a score lower by about one or two points.
pizza_cleaned %>%
filter(community_count >= 20) %>%
ggplot(aes(community_average_score, dave_average_score)) +
geom_point(aes(size = community_count)) +
geom_abline(color = "red") +
geom_smooth(method = "lm") +
labs(size = "# of community reviews",
x = "Community score",
y = "Dave score")