change colour ala gojek

2-clock.R

@@ -3,6 +3,7 @@
 # Email:           andi.herlan@protonmail.com
 # Data Used:       gojek.rds
 # Packages Used:   dplyr, lubridate, ggplot2, gridExtra
+# Source:          colours.R
 # Output File:     clock.png
 # Data Output:     -
 # Reference:       https://www.wjakethompson.com/post/2018-11-27-ggclock/
@@ -16,6 +17,9 @@ library(lubridate)
 library(ggplot2)
 library(gridExtra)
 
+# import colour definition
+source("colours.R")
+
 # main data
 gojek <- readRDS("output/gojek.rds")
 
@@ -56,10 +60,10 @@ p1 <- ggplot(data = clock) +
            label = paste0("Total\n", total_time, " hours\non the road")) +
   # ride
   geom_segment(aes(x, y, xend = x + 1, yend = y + 0.8), 
-               colour = "#00AA13", alpha = 0.4) +
+               colour = gc_green50, alpha = 0.4) +
   # car
   geom_point(data = filter(clock, clock$col == "Car"), 
-             aes(x, y), colour = "#000000", alpha = 0.8) +
+             aes(x, y), colour = "black", alpha = 0.8) +
   # coordinate
   coord_polar() +
   expand_limits(y = c(-1, 1)) +
@@ -68,7 +72,7 @@ p1 <- ggplot(data = clock) +
                      labels = c(3, 6, 9, "12")) +
   # text
   labs(title = "Time with GOJEK", x = NULL, y = NULL,
-       subtitle = paste("Andi's movement in", 
+       subtitle = paste("Movement in", 
                         min(year(gojek$datetime)), "-", 
                         max(year(gojek$datetime)), "(• is GoCar)", sep = " "),
        caption = paste0("Last movement at ", tail(gojek$datetime, 1),
@@ -97,7 +101,7 @@ p2 <- ggplot(data = dura) +
                 label = paste0("~", round(duration/3600), " hours")),
            size = 2.8, colour = "gray60", hjust = "left") +
   geom_linerange(aes(xmin = x0, xmax = x1, y = year),
-                 size = 4.6, alpha = 0.5, colour = "#00AA13") + 
+                 size = 4.6, alpha = 0.5, colour = gc_green50) + 
   scale_x_continuous(limits = c(0, limit_x_max)) +
   labs(title = "Duration per year",
        caption = "Github: akherlan | Data: GOJEK") +

3-distance.R

@@ -3,6 +3,7 @@
 # Email:           andi.herlan@protonmail.com
 # Data Used:       gojek.rds, java.rds
 # Packages Used:   dplyr, mapboxapi, ggplot2, ggrepel, gridExtra
+# Source:          colours.R
 # Output File:     distance.png
 # Data Output:     -
 # Reference:       https://ggplot2-book.org/annotations.html
@@ -18,6 +19,9 @@ library(ggplot2)
 library(ggrepel)
 library(gridExtra)
 
+# import colour definition
+source("colours.R")
+
 # main data
 gojek <- readRDS("output/gojek.rds")
 
@@ -76,7 +80,7 @@ p1 <- ggplot(data = distn) +
            colour = "gray50") +
   # late for the last train
   annotate(geom = "point", x = latetrain$duration/60, alpha = 0.3,
-           y = latetrain$distance, size = 4, col = "orange") +
+           y = latetrain$distance, size = 4, col = gc_orange) +
   annotate(geom = "curve", x = 50, y = 17, size = 0.3, col = "gray50",
            xend = latetrain$duration/60+0.6, yend = latetrain$distance,
            curvature = 0.3, arrow = arrow(length = unit(2, "mm"))) +
@@ -89,7 +93,7 @@ p1 <- ggplot(data = distn) +
            size = 3, col = "gray50") +
   # resettlement
   annotate(geom = "point", x = comeback$duration/60, alpha = 0.3,
-           y = comeback$distance, size = 4, col = "orange") +
+           y = comeback$distance, size = 4, col = gc_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"))) +
@@ -107,7 +111,7 @@ p1 <- ggplot(data = distn) +
   # 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")) +
+  scale_colour_manual(values = c("black", gc_green70)) +
   # main labels
   labs(
     title = "Movement with GOJEK",
@@ -134,7 +138,7 @@ p1 <- ggplot(data = distn) +
 # duration density
 dx <- ggplot(dens) +
   geom_density(aes(x = duration/60, y = -..density..), 
-               fill = "lightblue", colour = "gray35", alpha = 0.5) +
+               fill = gc_green20, colour = "gray35", alpha = 0.5) +
   labs(x = "Duration (minutes)", caption = "Github: akherlan | Data: GOJEK") +
   scale_x_continuous(limits = c(0, 75), breaks = c(seq(0, 15, 5), 30, 45, 60, 75)) +
   theme_minimal() +
@@ -154,7 +158,7 @@ dx <- ggplot(dens) +
 # distance density
 dy <- ggplot(dens) +
   geom_density(aes(x = -..density.., y = distance), 
-               fill = "lightblue", colour = "gray35", alpha = 0.5) +
+               fill = gc_green20, colour = "gray35", alpha = 0.5) +
   scale_y_continuous(breaks =  seq(0, 30, 5)) +
   labs(y = "Distance (km)", title = " ", subtitle = " ") +
   theme_minimal() +
@@ -178,7 +182,7 @@ p2 <- ggplot(java) +
   geom_sf(data = odline, colour = "black", size = 0.5, linetype = 2) +
   # location
   geom_point(data = od, aes(long, lat), 
-             colour = "red", size = 2, alpha = 0.6) +
+             colour = gc_red, size = 2, alpha = 0.6) +
   coord_sf() +
   theme_void()
 

4-cost.R

@@ -3,6 +3,7 @@
 # Email:           andi.herlan@protonmail.com
 # Data Used:       gojek.rds
 # Packages Used:   dplyr, tidyr, stringr, ggplot2
+# Source:          colours.R
 # Output File:     cost.png
 # Data Output:     -
 # Reference:       https://stackoverflow.com/questions/46597278/
@@ -11,17 +12,21 @@
 # clear environment
 rm(list = ls())
 
+# import colour definition
+source("colours.R")
+
 library(dplyr)
 library(tidyr)
 library(stringr)
 library(ggplot2)
 
 # main data
-gojek <- readRDS("output/gojek.rds")
+gojek <- readRDS("output/gojek.rds") %>% 
+  select(datetime, payment, price, discount, fee, paid)
 
-# data for cost
+# data for cost (rupiah in thousands)
 cost <- gojek %>% 
-  select(payment, price, discount, fee, paid) %>% 
+  select(-datetime) %>% 
   pivot_longer(cols = 2:5, names_to = "type", values_to = "rupiah") %>% 
   group_by(payment, type) %>% 
   summarise(rupiah = sum(rupiah/1000), .groups = "drop") %>% 
@@ -33,23 +38,21 @@ cost <- gojek %>%
 # all discount
 disc_total <- cost %>% group_by(type) %>% summarise(sum = sum(rupiah))
 
-# discount in app
+# discount, fee, price + fee in app
 godisc <- cost$rupiah[5]
+gofee <- cost$rupiah[6] 
+goprice <- round((cost$rupiah[8] + gofee)/1000, 2) # in millions IDR
 
-# price + fee in app
-fee <- cost$rupiah[6] # thousands
-goprice <- round((cost$rupiah[8] + fee)/1000, 2) # millions
-
-# percent cash expenses
+# percent cash expenses between cash and gopay
 paid_percent <- cost[cost$type == "Paid",] %>% 
   mutate(percent = round((rupiah/sum(rupiah))*100, 2))
 
-# percent cost (price + fee in both cahs and gopay)
+# percent cost (price + fee) in both cash and gopay
 cost_percent <- cost %>% 
   filter(cat == 1) %>% 
   mutate(percent = round((rupiah/sum(rupiah))*100, 2))
 
-# percent transactions done in app
+# transactions done in app and cash
 n_trans <- gojek %>% 
   select(paid, payment) %>% 
   group_by(payment) %>% 
@@ -59,7 +62,17 @@ n_trans <- gojek %>%
 # text postion transaction done in app
 y_trans <- (cost$rupiah[5] + cost$rupiah[7])*2/3
 
-# plot for costs
+# cost_growth <- gojek %>% 
+#   pivot_longer(cols = 3:6, names_to = "type", values_to = "rupiah") %>% 
+#   group_by(type) %>% 
+#   summarise(datetime = datetime, 
+#             cumsum = cumsum(rupiah), .groups = "drop") %>% 
+#   arrange(datetime)
+# 
+# ggplot(cost_growth) +
+#   geom_line(aes(datetime, cumsum, colour = type), show.legend = F)
+
+# plot for cost
 ggplot() + 
   # bar price + fee
   geom_bar(data = filter(cost, cat == 1), 
@@ -73,7 +86,8 @@ ggplot() +
   geom_text(data = filter(cost, cat == 1 & payment == "Cash"), 
             aes(x = 0, y = paid_percent$rupiah[1]/2, 
                 label = paste0(paid_percent$percent[1], "%\nof expenses")),
-            size = 4, colour = "#B3E7F3") +
+            # size = 4, colour = "#B3E7F3") +
+            size = 4, colour = "#DEF5FD") +
   # text price of all cost percent cash
   geom_text(data = filter(cost, cat == 1 & payment == "Cash"), 
             aes(x = 1, y = cost_percent$rupiah[2]/2, 
@@ -84,17 +98,19 @@ ggplot() +
             aes(x = 0, y = paid_percent$rupiah[2]/2, 
                 label = paste0(paid_percent$percent[2], "%\nof expenses\n(IDR ",
                                paid_percent$rupiah[2], ",000)")), 
-            size = 4, colour = "#B3E7F3") +
+            # size = 4, colour = "#B3E7F3") +
+            size = 4, colour = "#DEF5FD") +
   # text discount gopay
   geom_text(data = filter(cost, cat == 1 & payment == "GoPay"), 
             aes(x = 0, y = cost$rupiah[7] + cost$rupiah[5]/2, 
                 label = paste0("IDR ", godisc, ",000")), 
-            size = 4, colour = "#00BBE0") +
+            # size = 4, colour = "#00BBE0") +
+            size = 4, colour = "#388192") +
   # text price gopay
   geom_text(data = filter(cost, cat == 1 & payment == "GoPay"), 
             aes(x = 1, y = cost$rupiah[8]/2, 
                 label = paste0("Total costs\n IDR ", goprice, 
-                               " millions\n(include IDR ", fee, 
+                               " millions\n(include IDR ", gofee, 
                                ",000\nof additional app fee,\nparking, toll, etc.)\nwere paid inside app\nwith IDR ",
                                godisc, ",000\ndiscount+voucher")),
             size = 3.4, colour = "#005400") +
@@ -113,7 +129,8 @@ ggplot() +
        caption = "Github: akherlan | Data: GOJEK") +
   # styling
   facet_wrap(~payment, strip.position = "bottom") +
-  scale_fill_manual(values = c("#B3E7F3", "#000000", "#00BBE0", "#00AE00")) +
+  # scale_fill_manual(values = c("#B3E7F3", "#000000", "#00BBE0", "#00AE00")) +
+  scale_fill_manual(values = c("lightblue", "black", gc_blue, gc_green50)) +
   theme_minimal() +
   theme(
     text = element_text(family = "Sans"),
@@ -123,7 +140,7 @@ ggplot() +
     plot.subtitle = element_text(colour = "gray40"),
     plot.caption = element_text(colour = "gray60"),
     plot.title.position = "plot",
-    strip.text = element_text(face = "bold", size = 11),
+    strip.text = element_text(face = "bold", size = 11, colour = "gray40"),
     panel.grid.major.x = element_blank(),
     panel.grid.minor.x = element_blank(),
     panel.grid.minor.y = element_blank(),
@@ -136,4 +153,4 @@ ggplot() +
 
 # save PNG
 ggsave("cost.png", path = "figs", dpi = 150, units = "px",
-       width = 2*756, height = 2*509)
+       width = 2*600, height = 2*469)

5-needs.R

@@ -3,11 +3,11 @@
 # Email:           andi.herlan@protonmail.com
 # Data Used:       gojek.rds
 # Packages Used:   dplyr, tidyr, stringr, ggplot2, ggalluvial
+# Source:          colours.R
 # Output File:     needs.png
 # Data Output:     od.rds
 # Reference:       https://corybrunson.github.io/ggalluvial
 
-
 # clear environment
 rm(list = ls())
 
@@ -17,6 +17,9 @@ library(stringr)
 library(ggplot2)
 library(ggalluvial)
 
+# import colour definition
+source("colours.R")
+
 # main data
 gojek <- readRDS("output/gojek.rds")
 
@@ -92,9 +95,6 @@ od <- od %>%
                          !is.na(home), NA, TRUE))))
   )
 
-# empty category
-nrow(od) - colSums(!is.na(od[,4:8])) %>% as.matrix() %>% sum()
-
 # odplot <- od %>% 
 #   select(-datetime) %>%
 #   mutate(count = 1) %>% 
@@ -148,12 +148,13 @@ odp$destination <- odp$destination %>%
 # plot
 ggplot(data = odp,
        aes(axis1 = origin, axis2 = destination, y = freq)) +
-  scale_x_discrete(limits = c("Origin", "Needs"), 
-                   expand = c(0, 0.2, 0, 0.7)) +
   geom_alluvium(aes(fill = payment)) +
-  geom_stratum() +
+  geom_stratum(colour = "gray40") +
   geom_text(stat = "stratum", aes(label = after_stat(stratum))) +
-  scale_fill_manual(values = c("orange", "#00BBE0"),
+  scale_x_discrete(limits = c("Origin", "Needs"), 
+                   expand = c(0, 0.2, 0, 0.7)) +
+  # scale_fill_manual(values = c("orange", "#00BBE0"),
+  scale_fill_manual(values = c(gc_orange, gc_blue),
                     label = c("Cash", "GoPay")) +
   annotate(geom = "text", x = 2.25, y = 190, 
            label = paste("The most frequence trip", 
@@ -188,7 +189,6 @@ ggplot(data = odp,
     legend.position = c(0.75, 0.85),
     legend.direction = "vertical",
     legend.justification = "left",
-    # panel.border = element_rect(colour = "black", size = 1, fill = "transparent", linetype = 2),
     plot.title.position = "plot",
     plot.subtitle = element_text(colour = "gray40"),
     plot.caption = element_text(colour = "gray60"),

README.md

@@ -36,7 +36,7 @@ Biaya ini belum termasuk dengan penggunaan GoTagihan (GoBills) untuk BPJS, pulsa
 
 Perjalanan terbanyak ketika ng-Gojek adalah dari stasiun (atau halte) untuk pulang, disusul oleh perjalanan pulang selepas beraktivitas di luar selain kerja.
 
-Ada hubungan yang erat antara kantor dengan aktivitas transit.
+Ada hubungan yang erat antara kantor dengan aktivitas transit, dihubungkan dengan aktivitas ng-Gojek.
 
 Jika saya harus menggunakan uang tunai kemungkinan karena saya kehabisan GoPay saat berada di luar rumah.
 

colours.R

@@ -0,0 +1,20 @@
+# Title:           Colour Definition
+# Author:          Andi Herlan
+# Email:           andi.herlan@protonmail.com
+# Reference        https://www.gojek.com/data/
+
+# use source("colours.R") in any script for graphic generating
+
+"#006C0C" -> gc_green70     
+"#00AA13" -> gc_green50   # goride 
+"#4CC359" -> gc_green20     
+"#101820" -> gc_grayd     # dark
+"#727272" -> gc_grayl     # light-dark
+"#CACCCF" -> gc_black     
+"#E8E8E8" -> gc_black10   
+"#00AED6" -> gc_blue      # gopay
+"#772583" -> gc_purple    # gobiz
+"#BE1F2C" -> gc_red70     
+"#EE2737" -> gc_red       # gofood
+"#DF1995" -> gc_pink      # goplay
+"#F06400" -> gc_orange