-
Notifications
You must be signed in to change notification settings - Fork 3
/
forecast.R
108 lines (96 loc) · 3.76 KB
/
forecast.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
library(dplyr)
library(stringr)
library(readr)
library(tidyr)
library(ggplot2)
library(forcats)
recent_forcasts <- list.files("data_clean/", pattern = "_student.csv|_teacher.csv", full.names = TRUE) %>%
as_tibble() %>%
mutate(woche = str_extract(value, "[0-9]{6}")) %>%
arrange(desc(woche)) %>%
slice(1:2) %>%
pull(value)
forecast_students <- read_csv(grep("student", recent_forcasts, value = TRUE)) %>%
mutate(source = "student")
forecast_teacher <- read_csv(grep("teacher", recent_forcasts, value = TRUE)) %>%
mutate(source = "teacher")
time_kmk <- read_csv("data_clean/KMK_data.csv")
# combine both datasets
forecast <- bind_rows(forecast_students,
forecast_teacher)
get_country <- function(country_code, plot_string) {
#country_code <- "TH"
plot_data <- forecast %>%
mutate(key = paste0(type, "_", quantile),
ind_week = nchar(target_week),
target_week = ifelse(ind_week == 1, paste0("0", target_week), as.character(target_week)),
target_week = paste0(target_year, "-", target_week)) %>%
select(target_week, source, location, key, value) %>%
filter(location == country_code) %>%
mutate(value = as.numeric(value)) %>%
pivot_wider(names_from = key, values_from = value) %>%
bind_rows(time_kmk %>%
filter(location == country_code) %>%
mutate(ind_week = nchar(week),
target_week = ifelse(ind_week == 1, paste0("0", week), as.character(week)),
target_week = paste0(year, "-", target_week)) %>%
pivot_longer(cols = c('teacher', "student"), names_to = "source", values_to = "point_NA") %>%
select(target_week, source,
location, point_NA) %>%
mutate(point_NA = ifelse(point_NA == "-", NA, point_NA),
point_NA = as.numeric(point_NA)))
x_axis <- sort(plot_data$target_week)
plot <- plot_data %>%
mutate(source = ifelse(source == "student", "Schüler", "Lehrkräfte"),
source = fct_rev(source)) %>%
ggplot(aes(x = target_week, y = point_NA, group = 1)) +
geom_ribbon(
aes(ymin = quantile_0.025,
ymax = quantile_0.975),
alpha = 0.25,
fill = "#27408b"
) +
facet_wrap(~source, ncol = 1, scales = "free_y") +
theme_light(base_size = 30) +
theme(axis.text.x = element_text(angle = 45, vjust = 0.5, hjust = 0.5),
strip.background = element_rect(fill = "#152238"),
strip.text = element_text(colour = 'white', face = "bold")) +
scale_x_discrete(breaks = x_axis[seq(1, length(x_axis), 5)]) +
labs(x = "Kalenderwoche (KW)",
y = "Anzahl der Infizierten",
title = "Vorhersage infizierte Schüler und Lehrkräfte",
subtitle = paste0("Bundesland: ", plot_string)) +
geom_ribbon(aes(ymin = quantile_0.25,
ymax = quantile_0.75),
alpha = 0.25,
fill = "#27408b") +
geom_point(color = "#27408b", size = 3) +
geom_line(color = "#27408b", size = 1)
return(plot)
}
# get_country("BW", "Baden-Württemberg")
# get_country("TH", "Thüringen")
counties <- tribble(
~kurz, ~string,
"BW", "Baden-Württemberg",
"BY", "Bayern",
"BE", "Berlin",
"BB", "Brandenburg",
"HB", "Bremen",
"HH", "Hamburg",
"HE", "Hessen",
"NI", "Niedersachsen",
"MV", "Mecklenburg-Vorpommern",
"NW", "Nordrhein-Westfalen",
"RP", "Rheinland-Pfalz",
"SL", "Saarland",
"SN", "Sachsen",
"ST", "Sachsen-Anhalt",
"SH", "Schleswig-Holstein",
"TH", "Thüringen")
forecast_plots <- counties %>%
rowwise() %>%
mutate(plot = list(get_country(kurz, string)))
forecast_list <- forecast_plots$plot
names(forecast_list) <- forecast_plots$kurz
save(forecast_list, file = "leaflet_maps/forecast.RData")