COVID19 (Perú): Gráficos de positivos y fallecidos por semana epidemiológica

covid19-peru-positivos-fallecidos.R

# Inspirado en https://twitter.com/plenque/status/1475934854269259803

library(tidyverse)
library(lubridate)

pos_url <- "https://github.com/jmcastagnetto/covid-19-peru-limpiar-datos-minsa/raw/main/datos/positivos_covid_aumentado.rds"
pos <- rio::import(pos_url)

pos_semana <- pos %>%
  mutate(
    sunday_of_week = floor_date(
      fecha_resultado,
      "weeks",
      week_start = 7),
  ) %>%
  group_by(epi_year, epi_week, sunday_of_week) %>%
  tally(name = "positivos_semana")

fal_url <- "https://github.com/jmcastagnetto/covid-19-peru-limpiar-datos-minsa/raw/main/datos/fallecidos_covid_aumentado.rds"
fal <- rio::import(fal_url)

fal_semana <- fal %>%
  mutate(
    sunday_of_week = floor_date(
      fecha_fallecimiento,
      "weeks",
      week_start = 7),
  ) %>%
  group_by(epi_year, epi_week, sunday_of_week) %>%
  tally(name = "fallecidos_semana")

comb <- pos_semana %>%
  left_join(
    fal_semana,
    by = c("epi_year", "epi_week", "sunday_of_week")
  ) %>%
  filter(!is.na(epi_year)) %>%
  ungroup() %>%
  filter(sunday_of_week < max(sunday_of_week)) %>%
  mutate(
    yw = as.numeric(sprintf("%4d%02d", epi_year, epi_week)),
    yw_lbl = sprintf("%4d-S%02d", epi_year, epi_week)
  )

factor <- round(max(comb$positivos_semana) / max(comb$fallecidos_semana), -1)
min_yw <- comb %>%
  filter(yw == min(yw)) %>%
  pull(yw_lbl)
max_yw <- comb %>%
  filter(yw == max(yw)) %>%
  pull(yw_lbl)

comb <- comb %>%
  mutate(
    fallecidos_semana_scaled = factor * fallecidos_semana
  )

p1 <- ggplot(comb) +
  geom_line(aes(x = sunday_of_week, y = positivos_semana), color = "blue") +
  geom_line(aes(x = sunday_of_week, y = fallecidos_semana_scaled), color = "red") +
  scale_y_continuous(
    name = "Casos positivos",
    labels = scales::comma,
    sec.axis = sec_axis(
      name = "Fallecidos",
      trans = ~ . / factor,
      labels = scales::comma
    )
  ) +
  scale_x_date(
    name = "",
    date_labels = "%b\n%Y",
    date_breaks = "1 month"
  ) +
  labs(
    title = "COVID-19 (Perú): Casos positivos y fallecidos por semana epidemiológica",
    subtitle = glue::glue("Fuente: Datos abiertos del MINSA, de {min_yw} a {max_yw}"),
    caption = "@jmcastagnetto, Jesus M. Castagnetto"
  ) +
  theme_minimal(16) +
  theme(
    plot.background = element_rect(fill = "white"),
    axis.title.y.left = element_text(color = "blue", size = 18, face = "bold"),
    axis.text.y.left = element_text(color = "blue"),
    axis.title.y.right = element_text(color = "red", size = 18, face = "bold"),
    axis.text.y.right = element_text(color = "red"),
    plot.title.position = "plot",
    plot.title = element_text(size = 24),
    plot.subtitle = element_text(size = 18, color = "grey40"),
    plot.caption = element_text(family = "Inconsolata", size = 16),
    plot.margin = unit(rep(.5, 4), "cm")
  )

ggsave(
  p1,
  filename = "covid19-peru-positivos-fallecidos-semanaepi.png",
  height = 10,
  width = 14
)

p2 <- ggplot(comb) +
  geom_point(aes(x = sunday_of_week, y = positivos_semana), color = "blue") +
  geom_smooth(aes(x = sunday_of_week, y = positivos_semana), color = "blue", method = "gam") +
  geom_point(aes(x = sunday_of_week, y = fallecidos_semana_scaled), color = "red") +
  geom_smooth(aes(x = sunday_of_week, y = fallecidos_semana_scaled), color = "red", method = "gam") +
  scale_y_continuous(
    name = "Casos positivos",
    labels = scales::comma,
    sec.axis = sec_axis(
      name = "Fallecidos",
      trans = ~ . / factor,
      labels = scales::comma
    )
  ) +
  scale_x_date(
    name = "",
    date_labels = "%b\n%Y",
    date_breaks = "1 month"
  ) +
  labs(
    title = "COVID-19 (Perú): Casos positivos y fallecidos por semana epidemiológica",
    subtitle = glue::glue("Fuente: Datos abiertos del MINSA, de {min_yw} a {max_yw}"),
    caption = "Aproximación usando GAM [y ~ s(x, bs = 'cs')] - @jmcastagnetto, Jesus M. Castagnetto"
  ) +
  theme_minimal(16) +
  theme(
    plot.background = element_rect(fill = "white"),
    axis.title.y.left = element_text(color = "blue", size = 18, face = "bold"),
    axis.text.y.left = element_text(color = "blue"),
    axis.title.y.right = element_text(color = "red", size = 18, face = "bold"),
    axis.text.y.right = element_text(color = "red"),
    plot.title.position = "plot",
    plot.title = element_text(size = 24),
    plot.subtitle = element_text(size = 18, color = "grey40"),
    plot.caption = element_text(family = "Inconsolata", size = 16),
    plot.margin = unit(rep(.5, 4), "cm")
  )

ggsave(
  p2,
  filename = "covid19-peru-positivos-fallecidos-semanaepi-gam.png",
  height = 10,
  width = 14
)

Gráfico de líneas

Gráfico con aproximación usando GAM