spotify_all_years_YlGn5.R

library(jsonlite)
library(tidyverse)
library(RColorBrewer)
library(ggtext)
library(OneR)

#extrafont::loadfonts() # fix fonts

streamHistory <-
  list.files(
    path = "ext/",
    full.names = TRUE,
    pattern = ".json"
  ) %>%
  map_df(~ fromJSON(.)) %>%
  filter(ms_played / 1000 > 30 & !duplicated(.)) %>%
  as_tibble() %>%
  mutate_at("ts", ymd_hms, tz = "Asia/Yekaterinburg") %>%
  mutate(
    date = floor_date(ts, "day") %>% as_date(),
    day = day(ts),
    month = month(ts),
    year = year(ts)
  ) %>%
  filter(year %in% c(2019:2023))

# str(streamHistory)
# names(streamHistory)

tracks_count <- streamHistory %>%
  group_by(year, date) %>%
  count() %>%
  ungroup() %>%
  mutate(year_day = yday(date))

write_csv2(tracks_count, "tracks_count.csv")
# str(tracks_count)

quantile(tracks_count$n, probs = seq(0, 1, 0.1))
quantile(tracks_count$n, probs = seq(0, 1, 0.25))
table(bin(tracks_count$n, nbins = 5, method = "content"))
table(
  bin(tracks_count$n, nbins = 5, method = "content"),
  tracks_count$year
)

# col_strip <- brewer.pal(n = 5, "Oranges")
col_strip <- brewer.pal(n = 5, "YlGn")

# quantiles
q <- quantile(tracks_count$n, seq(0, 1, 0.2))
qr <- quantile(rank(tracks_count$n), seq(0, 1, 0.2))
# attributes(q)$names[2:5]

labelsq <- paste0(
  as.numeric(q), "\n",
  c("min", attributes(q)$names[2:5], "max")
)

all_tracks <- sum(tracks_count$n)

median <- median(tracks_count$n)

tracks_by_year <- tracks_count %>% 
  group_by(year) %>%
  summarise(sum = sum(n), 
            median = median(n))
tracks_by_year

annot <- glue::glue("Учитывались треки с продолжительностью прослушивания 
                    более 30 секунд. Всего было прослушано {all_tracks} 
                    треков за весь период с 2019 по 2023 год. Медианное 
                    значение числа ежедневного прослушивания 
                    составляет {median} треков")
annot <- stringr::str_wrap(annot, 68)

tracks_count %>%
  ggplot(aes(
    x = year_day, y = 1,
    fill = rank(n)
  )) +
  geom_tile() +
  scale_fill_stepsn(
    colours = col_strip,
    breaks = qr,
    labels = labelsq
  ) +
  facet_wrap(~year,
    ncol = 1,
    strip.position = "left"
  ) +
  labs(
    x = "",
    y = "",
    caption = "Данные: Аккаунт пользователя Spotify<br>
Юрий Тукачев, ноябрь 2023 @weekly_charts"
  ) +
  ggtitle(
    "Как часто я слушал музыку на Spotify?",
    paste0("Ежедневное число прослушиваний треков в течение года\n", annot)
  ) +
  guides(
    fill = guide_coloursteps(
      title = "Число треков",
      title.theme = element_text(
        size = 16,
        margin = margin(t = 15)
      ),
      label.theme = element_text(
        color = "grey30",
        size = 14,
        lineheight = 1.1
      ),
      title.position = "top",
      title.hjust = .5,
      barwidth = unit(15, "lines"),
      barheight = unit(.5, "lines")
    )
  ) +
  theme_void() +
  theme(
    plot.background = element_rect(color = "grey60"),
    plot.title = element_markdown(face = "bold"),
    plot.title.position = "plot",
    text = element_text(
      size = 22,
      family = "PT Sans"
    ),
    plot.caption.position = "plot",
    plot.caption = element_markdown(
      lineheight = 1.1,
      size = 16,
      color = "gray60",
      margin = margin(t = 20)
    ),
    plot.subtitle = element_text(
      size = 16,
      color = "gray60",
      margin = margin(t = 15, b = 20)
    ),
    legend.position = "bottom",
    panel.background = element_blank(),
    plot.margin = margin(25, 25, 10, 25),
    axis.ticks = element_blank()
  )

ggsave(
  "spotify_all_years_YlGn5.png",
  bg = "white",
  dpi = 600,
  width = 8,
  height = 8
)