Recursively maps intervals to non-overlapping segments

packed-interval-chart.R

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


## takes playlist list and returns min and max track_added_at for each playlist:
endpoints <- lol %>%
  filter(playlist_owner == user_id) %>%
  mutate(track_added_at = as.Date(track_added_at)) %>%
  filter(track_added_at <=  Sys.Date()) %>%
  select(playlist_id, playlist_name, track_added_at) %>%
  group_by(playlist_id, playlist_name) %>%
  summarise(n_tracks = n(), start = min(track_added_at), end = max(track_added_at)) %>%
  ungroup()

endpoints <- filter(endpoints, year(start) >= 2012)

## Initializing the recursive process. Takes a data frame of intervals, and sorts them into groups within which the intervals don't overlap.
## starts with the first interval, finds the next earliest interval whose start is greater that the first interval's endpoint.
## repeats process until we get to last non-overlapping interval.
## increments position, and starts with earliest remaining interval...

## Starts with a, a dataframe we continue to remove the first element from.
# a requires rows that contain intervals based on the variables start and end
a <- arrange(endpoints, start)

## empty dataset with the same fields as a, output will be same rows, with an additional 
output <- a %>% filter(FALSE)
output <- output %>% mutate(position = 0)

# counter that maps to output$position
pos = 0

# Recursive function:
place_item <- function(x) {
  if (nrow(a) > 0) {
    ## Take the first row, of 
    b <- filter(x, row_number() == 1) %>% mutate(position = pos)
    
    ## map row to the output
    output <<-  bind_rows(output, b)
    
    
    a <<- filter(a, playlist_id != b$playlist_id)
    
    # takes all intervals
    c <- filter(a, start > max(b$end))
    
    if (nrow(c) > 0) {
      place_item(c)
    }
    {
    pos <<- pos + 1
    place_item(a)
    }
  }
}

# run the recursion
x <- place_item(a)


theme_void_alt <- function (base_size = 12, base_family = "") 
{
  theme(rect = element_blank(), text = element_text(family = base_family, 
                                                    face = "plain", colour = "black", size = base_size, lineheight = 0.9, 
                                                    hjust = 0.5, vjust = 0.5, angle = 0, margin = margin(), 
                                                    debug = FALSE), plot.margin = unit(c(0, 0, 0, 0), "lines"), 
        axis.text.x = element_blank(), axis.text.y = element_blank(), 
        axis.title.x = element_blank(), axis.title.y = element_blank(), 
        legend.text = element_text(size = rel(0.8)), legend.title = element_blank(),
        axis.ticks = element_blank(),
        strip.text = element_text(size = rel(0.8)), complete = TRUE)
}        

output %>%
  ggplot(aes(x = -position, ymin = start, ymax = end)) + geom_linerange(size = 5) +
  coord_flip() +
  theme_void() + theme(axis.text.x = element_text(), plot.margin = margin(0, 0, 10, 10))

output_b <- output %>%
  mutate(diff = end - start) %>%
  arrange(desc(diff)) %>%
  mutate(i = row_number(),
         label = ifelse(i <= 5, playlist_name, ""),
         col = ifelse(i <= 10, -i, -20),
         middate = start + diff/2) 

output_b %>%
  filter(position <= 8) %>%
  ggplot(aes(x = -position, ymin = start, ymax = end+1,y=middate, label = label)) + geom_linerange(aes(color = col),size = 7) +
  coord_flip() + geom_text(color =  viridis(1), fontface = 'bold') +
  scale_color_viridis() +
  theme_void_alt() +
  theme(axis.text.x = element_text(),
        plot.margin = margin(0, 0, 10, 10),
        legend.position = 'none',
        panel.grid.major.x = element_line(color = "gray")
  ) + ggtitle("X's Spotify Playlists", "Duration from playlist creation to last update")

# 
# output_b %>%
#   filter(position <= 8) %>%
#   ggplot(aes(x = -position, ymin = start, ymax = end+1,y=middate, label = label)) + geom_linerange(aes(color = col),size = 7) +
#   coord_flip() + geom_text(color =  viridis(1), fontface = 'bold') +
#   scale_color_viridis() +
#   theme_void_alt() +
#   theme(axis.text.x = element_text(),
#         plot.margin = margin(0, 0, 10, 10),
#         legend.position = 'none',
#         panel.grid.major.x = element_line(color = "gray") + ggtitle("Alex's Spotify Playlists")

# lol %>%
#   ggplot(aes(x=as.Date(track_added_at))) + geom_density()
# 

#lol <- data.frame(do.call(rbind, list_of_lists))
#lol$track_artists_mess[2][[1]]

filter(lol, playlist_owner = user_id)