include data and add a map to 'distance' graph

1-wrangling.R

@@ -0,0 +1,167 @@
+# Title:           Data Wrangling
+# Author:          Andi Herlan
+# Email:           andi.herlan@protonmail.com
+# Data Used:       travel.csv
+# Packages Used:   stringr, lubridate, dplyr, tidyr
+# Output File:     -
+# Data Output:     gojek.rds
+
+library(dplyr)
+library(tidyr)
+library(stringr)
+library(lubridate)
+
+
+# invoice mail from GOJEK
+travel <- read.csv("travel.csv")
+content <- travel$body
+gojek <- travel %>% as_tibble()
+
+# datetime order == datetime email
+gojek$datetime <- as_datetime(gojek$datetime, 
+                              format = "%m/%d/%Y %H:%M:%S", 
+                              tz = "Asia/Jakarta")
+
+# ride or car
+gojek$vehicle <- content %>% 
+  str_extract("G[O|o]-?\\w{3,4}") %>% 
+  str_to_lower() %>% 
+  str_remove("go-?") %>% 
+  str_to_title()
+
+# distance
+gojek$distance <- content %>% 
+  str_extract("[D|J]\\w+.+\\skm") %>% 
+  str_extract("\\d+(\\.\\d+)?") %>% 
+  as.numeric()
+
+# duration
+dur <- content %>% 
+  str_extract("[T|W]\\w+\\s\\w+\\s(:?\\d{2}){3}") %>% 
+  str_extract("(:?\\d{2}){3}") %>% 
+  as_tibble() %>% 
+  separate(value, sep = ":", into = c("h", "m", "s"))
+
+dur <- sapply(dur, as.numeric) %>% as_tibble()
+dur$duration <- dur$h*60*60 + dur$m*60 + dur$s
+gojek$duration <- duration(dur$duration)
+
+# price
+gojek$price <- content %>% 
+  str_extract("\\w+\\s\\(.+\\)\\sRp\\d+\\.\\d+") %>% 
+  str_extract("Rp\\d+(\\.\\d+)?") %>% 
+  str_remove("Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# discount
+gojek$discount <- content %>% 
+  str_extract("Diskon.+") %>% 
+  str_remove("Diskon.+Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# voucher
+gojek$voucher <- content %>%
+  str_extract("Voucher.+") %>% 
+  str_remove("Voucher.+Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# app service fee
+gojek$fee <- content %>% 
+  str_extract("Biaya jasa aplikasi.+\\d{3}") %>% 
+  str_remove("^B.+Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# additional fee
+gojek$additional <- content %>% 
+  str_extract("Pendapatan tambahan.+") %>% 
+  str_remove("^P.+Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# toll and parking
+gojek$toll <- content %>% 
+  str_extract("Ongkos tol/parkir.+") %>% 
+  str_remove("^O.+Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# total payment
+gojek$paid <- content %>% 
+  str_extract("TOTAL.+") %>% 
+  str_extract("Rp\\d+(\\.\\d+)?") %>% 
+  str_remove("Rp") %>% 
+  str_remove("\\.") %>% 
+  as.numeric()
+
+# payment method
+gojek$payment <- content %>% 
+  str_extract("G[O|o]-?P.+") %>% 
+  str_remove("-") %>% 
+  str_to_title() %>% 
+  str_trim()
+
+gojek <- gojek %>% 
+  mutate(payment = ifelse(is.na(payment), "Cash", payment))
+
+# pickup
+gojek$pickup <- content %>% 
+  str_squish() %>% 
+  str_extract("image: pickup.+image: drop") %>% 
+  str_remove_all("image:") %>% 
+  str_remove("pickup") %>% 
+  str_remove("drop") %>% 
+  str_remove("\\]") %>% 
+  str_remove("\\[") %>% 
+  str_remove("  ") %>% 
+  str_remove("Penjemputan\\s?\\*\\s?\\•\\s?\\d{2}:\\d{2}\\s?\\*") %>% 
+  str_remove("pick up\\s?\\*\\s?\\•\\s?\\d{2}:\\d{2}\\s?\\*") %>% 
+  str_trim()
+
+# destination
+gojek$destination <- content %>% 
+  str_squish() %>% 
+  str_extract("image: drop.+image: Driver Image") %>% 
+  str_remove_all("[I|i]mage:?") %>% 
+  str_remove("drop") %>% 
+  str_remove("\\]") %>% 
+  str_remove("\\[") %>% 
+  str_remove("Driver") %>% 
+  str_remove("   ") %>% 
+  str_remove("Tujuan\\s?\\*\\s?\\•\\s?\\d{2}:\\d{2}\\s?\\*") %>% 
+  str_remove("destination\\s?\\*\\s?\\•\\s?\\d{2}:\\d{2}\\s?\\*") %>% 
+  str_trim()
+
+# driver
+gojek$driver <- content %>% 
+  str_extract("Driver Image\\]\\s\n\n.+") %>%
+  str_remove_all("Driver Image] \n\n") %>% 
+  str_remove("(Your driver)?(Driver Anda)?") %>% 
+  str_trim()
+
+gojek <- gojek %>% select(-2:-4)
+
+# add voucher to discount; gather fees; NA == 0
+gojek <- gojek %>% 
+  mutate(discount = ifelse(is.na(discount), 0, discount),
+         voucher = ifelse(is.na(voucher), 0, voucher),
+         discount = discount + voucher,
+         fee = ifelse(is.na(fee), 0, fee),
+         additional = ifelse(is.na(additional), 0, additional),
+         toll = ifelse(is.na(toll), 0, toll),
+         fee = fee + additional + toll) %>% 
+  select(-voucher, -additional, -toll)
+
+
+# check: price - discount + fee = paid
+gojek$price - gojek$discount + gojek$fee == gojek$paid
+
+# total consumption
+colSums(gojek[,c(3:8)])
+
+# save to RDS
+saveRDS(gojek, file = "output/gojek.rds")
+

2-clock.R

@@ -0,0 +1,75 @@
+# Title:           Clock
+# Author:          Andi Herlan
+# Email:           andi.herlan@protonmail.com
+# Data Used:       gojek.rds
+# Packages Used:   dplyr, lubridate, ggplot2
+# Output File:     clock.png
+# Data Output:     -
+# Reference:       https://www.wjakethompson.com/post/2018-11-27-ggclock/
+
+
+# clear environment
+rm(list = ls())
+
+library(dplyr)
+library(lubridate)
+library(ggplot2)
+
+# main data
+gojek <- readRDS("output/gojek.rds")
+
+# data for clocking
+clock <- gojek %>% 
+  summarise(datetime = datetime) %>% 
+  mutate(y = as.numeric(pm(datetime)),
+         x = hour(datetime),
+         x = ifelse(y == 1, x-12L, x)*60L,
+         x = x + minute(datetime),
+         col = gojek$vehicle)
+
+# plot
+ggplot(data = clock) +
+  # am / pm
+  annotate(geom = "text", x = 0, y = 0.5, label = "AM",
+           size = 4, colour = "gray40", alpha = 0.3) +
+  annotate(geom = "text", x = 0, y = 1.5, label = "PM",
+           size = 4, colour = "gray40", alpha = 0.3) +
+  # ride
+  geom_segment(aes(x, y, xend = x + 1, yend = y + 0.8), 
+               colour = "#00AA13", alpha = 0.4) +
+  # car
+  geom_point(data = filter(clock, clock$col == "Car"), 
+             aes(x, y), colour = "#000000", alpha = 0.8) +
+  # coordinate
+  coord_polar() +
+  expand_limits(y = c(-1, 1)) +
+  scale_x_continuous(limits = c(0, 720),
+                     breaks = seq(180, 720, 180), 
+                     labels = c(3, 6, 9, "12")) +
+  # text
+  labs(title = "Travel with GOJEK", x = NULL, y = NULL,
+       subtitle = paste("Andi's movement in", 
+                        min(year(gojek$datetime)), "-", 
+                        max(year(gojek$datetime)), "(• is GoCar)", sep = " "),
+       caption = paste("Last movement at", tail(gojek$datetime, 1),
+                       "AM for vaccination\nGithub: akherlan | Data: GOJEK", 
+                       sep = " ")) +
+  # styling
+  theme_minimal() +
+  theme(
+    text = element_text(family = "Sans"),
+    axis.title = element_blank(),
+    axis.ticks = element_blank(),
+    axis.text.x = element_text(size = 15),
+    axis.text.y = element_blank(),
+    panel.grid.major.x = element_line(size = 0.3),
+    panel.grid.major.y = element_line(size = 0.3, linetype = 2),
+    panel.grid.minor = element_blank(),
+    plot.subtitle = element_text(colour = "gray40"),
+    plot.caption = element_text(colour = "gray60"),
+    plot.background = element_rect(fill = "white", size = 0)
+  )
+
+# save PNG
+ggsave("clock.png", path = "figs", dpi = 150, units = "px",
+       width = 2*540, height = 2*507)

3-distance.R

@@ -0,0 +1,116 @@
+# Title:           Distance
+# Author:          Andi Herlan
+# Email:           andi.herlan@protonmail.com
+# Data Used:       gojek.rds
+# Packages Used:   dplyr, mapboxapi, ggplot2, ggrepel
+# Output File:     distance.png
+# Data Output:     -
+# Reference:       https://ggplot2-book.org/annotations.html
+
+
+# clear environment
+rm(list = ls())
+
+library(dplyr)
+library(mapboxapi)
+library(ggplot2)
+library(ggrepel)
+
+# main data
+gojek <- readRDS("output/gojek.rds")
+
+# data for distance 
+distn <- gojek %>% 
+  select(vehicle, distance, duration)
+
+# resettlement
+comeback <- gojek %>% 
+  arrange(distance) %>% 
+  tail(1)
+
+# aladdin's carpet
+wrongride <- gojek %>% 
+  arrange(duration) %>% 
+  head(3)
+
+# plot 
+p1 <- ggplot(data = distn) +
+  # total distance
+  annotate(geom = "text", x = 1.8, y = 28, size = 3.2, hjust = "left",
+           label = paste0("Total distance:\n", sum(gojek$distance),
+                         "km (", nrow(gojek), " trips)"),
+           colour = "gray50") +
+  # total duration
+  annotate(geom = "text", x = 56, y = 2, size = 3.2, hjust = "right",
+           label = paste0(round(sum(gojek$duration)/3600, 2), 
+                          " hours\non the road"),
+           colour = "gray50") +
+  # resettlement
+  annotate(geom = "point", x = comeback$duration/60, alpha = 0.4,
+           y = comeback$distance, size = 3.5, col = "orange") +
+  annotate(geom = "curve", x = 62, y = 25, size = 0.3, col = "gray50",
+           xend = comeback$duration/60-0.2, yend = comeback$distance-0.4,
+           curvature = 0.3, arrow = arrow(length = unit(2, "mm"))) +
+  annotate(geom = "text", label = "Resettled\nfrom Srengseng\nto Depok", 
+           x = 61.4, y = 25, hjust = "right", size = 3.5, col = "gray50") +
+  # must be wrong
+  annotate(geom = "curve", x = 10.4, y = 15.5, size = 0.3, col = "gray50",
+           xend = wrongride$duration[1]/60+0.2, yend = wrongride$distance[1]+0.4,
+           curvature = 0.3, arrow = arrow(length = unit(2, "mm"))) +
+  annotate(geom = "text", label = "Aladdin's carpet\n(only ~3-6 second!)", 
+           hjust = "left", x = 11.4, y = 15.5, size = 3, col = "gray50") +
+  # scatter plot
+  geom_point(aes(duration/60, distance, colour = vehicle), alpha = 0.4) +
+  # scales
+  scale_y_continuous(breaks =  seq(0, 30, 5)) +
+  scale_x_continuous(limits = c(0, 75), breaks = c(seq(0, 15, 5), 30, 45, 60, 75)) +
+  scale_colour_manual(values = c("#000000", "#00AA13")) +
+  # main labels
+  labs(title = "Travel with GOJEK", 
+       subtitle = paste0("Equivalent to Jakarta-Surabaya (781 km) by car"),
+       caption = "Github: akherlan | Data: GOJEK",
+       x = "Duration (minutes)", y = "Distance (km)",
+       colour = "GO") +
+  # styling
+  theme_minimal() +
+  theme(
+    text = element_text(family = "Sans"),
+    legend.position = c(0.85, 0.28),
+    panel.grid.minor = element_blank(),
+    plot.subtitle = element_text(colour = "gray40"),
+    plot.caption = element_text(colour = "gray60"),
+    axis.title = element_text(colour = "gray50"),
+    axis.text = element_text(colour = "gray50"),
+    plot.background = element_rect(fill = "white", size = 0)
+  )
+
+# save PNG
+ggsave("distance.png", path = "figs", dpi = 150, units = "px",
+       width = 2*817, height = 2*516)
+
+# add map data
+java <- readRDS("output/java.rds")
+
+# geocoding
+js <- tribble(~city, "Jakarta", "Surabaya")
+jscoord <- lapply(js$city, mb_geocode) %>% as.data.frame()
+jscoord <- as_tibble(t(jscoord))
+names(jscoord) <- c("long", "lat")
+js <- bind_cols(js, jscoord)
+
+# longtrip Jakarta - Surabaya
+jsline <- mb_directions(origin = "Jakarta", destination = "Surabaya")
+
+p2 <- ggplot(java) +
+  geom_sf(fill = "gray90", alpha = 0.3, colour = "gray80", size = 0.2) +
+  geom_sf(data = jsline, colour = "black", size = 0.5, linetype = 2) +
+  geom_point(data = js, aes(long, lat), 
+             colour = "red", size = 2, alpha = 0.6) +
+  coord_sf() +
+  theme_void()
+
+p1 + annotation_custom(ggplotGrob(p2), xmin = 3, xmax = 45, ymin = 18, ymax = 27)
+
+# save PNG
+ggsave("distance_map.png", path = "figs", dpi = 150, units = "px",
+       width = 2*817, height = 2*516)