covproj.R

# Load packages -----------------------------------------------------------

library(httr)       # CRAN v1.4.2
library(dplyr)      # CRAN v1.0.1
library(lubridate)  # CRAN v1.7.9
library(tidyr)      # CRAN v1.1.1
library(zoo)        # CRAN v1.8-8
library(ggplot2)    # CRAN v3.3.2
library(hrbrthemes) # CRAN v0.8.0

# Fetch COVID-19 data in regional level -----------------------------------

#' In this example we will inspect Jawa Barat data. Please pick one province according to your interest, other than Jawa Barat, to be used as case study :)

resp <- GET("https://data.covid19.go.id/public/api/prov_detail_JAWA_BARAT.json")
status_code(resp)

cov_prov_raw <- content(resp, as = "parsed", simplifyVector = TRUE)
str(cov_prov_raw, max.level = 2, vec.len = 3)

# Inspect the daily cases, pre-process data if necessary ------------------

cov_prov_daily <-
  cov_prov_raw$list_perkembangan %>%
  transmute(
    date = as.POSIXct(tanggal / 1000, origin = "1970-01-01") %>%
      as.Date(),
    newcase = KASUS,
    recovered = SEMBUH,
    death = MENINGGAL
  )

glimpse(cov_prov_daily)

# Turn table into graph! --------------------------------------------------

cov_prov_daily %>%
  ggplot(aes(date, newcase)) +
  geom_col(fill = "firebrick3") +
  scale_x_date(
    breaks = "2 weeks",
    guide = guide_axis(check.overlap = TRUE, n.dodge = 2),
    labels = scales::label_date(format = "%e %b"),
    expand = c(0.005, 0.005)
  ) +
  labs(
    x = NULL,
    y = "Jumlah kasus",
    title = "Kasus Harian Positif COVID-19 di Jawa Barat",
    subtitle = "Terjadi pelonjakan kasus di awal bulan Juli akibat klaster Secapa AD Bandung",
    caption = "Sumber data: covid.19.go.id"
  ) +
  theme_ipsum_tw(
    base_size = 13,
    plot_title_size = 21,
    grid = "Y",
    ticks = TRUE
  ) +
  theme(plot.title.position = "plot")

#' Please continue with your own for the other variables! Do not forget to draw some conclusion about them :)

# Daily cases, rolling 3-day average --------------------------------------

#' This is a bonus section for you! Please forgive me for giving you such a lengthy code below :p

cov_prov_daily %>%
  mutate(
    across(newcase:death, ~ rollmean(.x, k = 3, fill = NA))
  ) %>%
  pivot_longer(
    cols = newcase:death,
    names_to = "status",
    values_to = "rollmean3day"
  ) %>%
  mutate(
    status = factor(status, levels = c("newcase", "recovered", "death"), labels = c("Positif", "Sembuh", "Meninggal"))
  ) %>%
  ggplot(aes(date, rollmean3day, colour = status)) +
  facet_wrap(~status, ncol = 1, scales = "free_y") +
  geom_line(size = 1.1, show.legend = FALSE) +
  scale_x_date(
    breaks = "10 days",
    guide = guide_axis(check.overlap = TRUE, n.dodge = 2),
    labels = scales::label_date(format = "%e %b"),
    expand = c(0.005, 0.005)
  ) +
  scale_y_continuous(position = "right") +
  scale_colour_manual(
    values = c("firebrick3", "seagreen2", "darkslategray4")
  ) +
  labs(
    x = NULL,
    y = NULL,
    title = "Kasus Harian COVID-19 di Jawa Barat Menggunakan Rerata Bergerak 3 hari",
    subtitle = "Jumlah kasus pada sumbu-y tidak dibuat seragam antar panel status kasus",
    caption = "Sumber data: covid.19.go.id"
  ) +
  theme_ipsum_tw(
    base_size = 13,
    plot_title_size = 21,
    strip_text_face = "italic",
    grid = FALSE,
    ticks = TRUE
  ) +
  theme(plot.title.position = "plot")

#' What can you learn from the graph above? Also, do not hesitate to ping me at Telegram if you want to confirm some code snippets (please find my username at the very end of this document).

# Transform daily cases into weekly cases ---------------------------------

cov_prov_weekly <-
  cov_prov_daily %>%
  group_by(
    year = year(date),
    week = week(date)
  ) %>%
  summarise(
    across(c(newcase:death), ~ sum(.x, na.rm = TRUE))
  ) %>%
  ungroup()

glimpse(cov_prov_weekly)


                    # [Again] turn table into graph! ------------------------------------------

#' You can go wild with your creativity here, tell me what you learn from the graphs!

# Is this week is better than last week? ----------------------------------

cov_prov_weekly_comparison <-
  cov_prov_weekly %>%
  transmute(
    year,
    week,
    newcase,
    newcase_lastweek = dplyr::lag(newcase, 1), # can you explain why we use of `::` operator here?
    newcase_lastweek = replace_na(newcase_lastweek, 0),
    is_better = newcase < newcase_lastweek
  )
glimpse(cov_prov_weekly_comparison)

cov_prov_weekly_comparison %>%
  count(is_better)

#' Please put your thoughts into this graph!

cov_prov_weekly_comparison %>%
  ggplot(aes(week, newcase, fill = is_better)) +
  geom_col(show.legend = FALSE) +
  scale_x_continuous(expand = c(0, 0)) +
  scale_fill_manual(values = c("TRUE" = "seagreen3", "FALSE" = "salmon")) +
  labs(
    x = NULL,
    y = "Jumlah kasus",
    title = "Kasus Pekanan Positif COVID-19 di Jawa Barat",
    subtitle = "Kolom hijau menunjukan penambahan kasus baru lebih sedikit dibandingkan satu pekan sebelumnya",
    caption = "Sumber data: covid.19.go.id"
  ) +
  theme_ipsum_tw(
    base_size = 13,
    plot_title_size = 21,
    grid = "Y",
    ticks = TRUE
  ) +
  theme(plot.title.position = "plot")

#' I challenge you to make the similar graph for the recovered and death cases :)

# Is this week is better than last week and the last two week? ------------

#' Using the same principles as in the previous section, now it's your turn to shine! Please highlight some conclusion about your findings!

# The Seventh-day Amplification Factor ------------------------------------

#' Please consult to these documents:
#' * https://www.bangkokpost.com/opinion/opinion/1902320/taking-the-fight-to-covid-19-using-simple-mathematics
#' * https://towardsdatascience.com/how-the-seventh-day-amplification-factor-can-gauge-the-existence-of-corona-silent-carriers-b9414ef3df62
#'
#' Again, using the similar approach as in the previous section and from what you've learn from DQLab project, I believe that you can calculate the Seventh-day Amplification Factor easily!
#' Here are some clues that I have for you (in no particular order):
#' * cov_prov_daily
#' * cumsum()
#' * 3-day rolling average
#' * lag()
#' Please take those hints to calculate the Seventh-day Amplification Factor for your province of interest!
#'
#' When you are finished, please share your finding in LinkedIn and tag both DQLab and me. We eagerly looking forward to it!
#'
#'
#' Kind regards,
#'
#' Muhammad Aswan Syahputra
#' linkedin.com/in/aswansyahputra
#' t.me/aswansyahputra
#' github.com/aswansyahputra

saya sudah mencobanya terima kasih

Wah sama-sama mbak @utamiprtw. Bagaimana apakah berhasil?

Alhamdulillah berhasil kak

…

On Fri, 11 Sep 2020 at 10.09 Muhammad Aswan Syahputra < ***@***.***> wrote: ***@***.**** commented on this gist. ------------------------------ Wah sama-sama mbak. Bagaimana apakah berhasil? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <https://gist.github.com/acc249f6ad716ae007f8879c0ad8f353#gistcomment-3449901>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQ6J4DUWONPEDWEFAE7LVXDSFGIH5ANCNFSM4RGBBB7Q> .

Terimakasih banyak atas bimbingannya pak Aswan, code yang terlampir sangat membantu pemula yang sedang belajar

terimakasih sangat membantu

kenapa %>% saya belum aktif juga ya pak?

kenapa %>% saya belum aktif juga ya pak?

Apakah paket {magrittr} atau {tidyverse} sudah diaktifkan @chan-tik?

Alhamdulillah, sama-sama @argavi1 @gitanadiavio