-
Notifications
You must be signed in to change notification settings - Fork 3
/
two_covid_canadas.Rmd
254 lines (222 loc) · 11.5 KB
/
two_covid_canadas.Rmd
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
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
---
title: "Two Covid Canadas"
author: "Jens von Bergmann"
date: "Last updated at `r format(Sys.time(), '%d %B, %Y - %H:%M',tz='America/Vancouver')`"
output: rmarkdown::github_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(
echo = FALSE,
message = FALSE,
warning = FALSE,
fig.retina = 2,
dpi = 150,
fig.width = 7,
fig.height = 5
)
library(ggplot2)
library(dplyr)
library(tidyr)
library(here)
library(cansim)
library(CanCovidData)
source(here("R/helpers.R"))
```
This notebook has been discontinued
This notebook shows the confirmed COVID cases for Canadian provinces. The code for this notebook is [available for anyone to adapt and use for their own purposes](https://github.com/mountainMath/BCCovidSnippets/blob/main/two_covid_canadas.Rmd).
```{r}
pop_data <- simpleCache(get_cansim("17-10-0005") %>%
filter(REF_DATE==2020,`Age group`=="All ages",Sex=="Both sexes") %>%
select(Province=GEO,Population=VALUE),"prov_pop_data",path = here::here("data"))
covid_data <- get_canada_official_provincial_data() %>%
mutate(shortProvince=recode(shortProvince,"Nouveau-Brunswick"="NB")) %>%
mutate(Province=recode(prname,"Nouveau-Brunswick"="New Brunswick")) %>%
filter(!(Province %in% c("Canada","Repatriated"))) %>%
mutate(update=coalesce(update,FALSE))
successful_provinces <- c("NL","NT","NS","YT","PE","NB")
province_colours <- c(setNames(RColorBrewer::brewer.pal(6,"Dark2"),c("SK","AB","BC","MB","ON","QC")),
setNames(RColorBrewer::brewer.pal(4,"Dark2"),c("NL","NS","PE","NB")))
successful_label <- "Atlantic provinces"
successful_provinces <- c("NT","NS","YT","PE","NB")
province_colours <- c(setNames(RColorBrewer::brewer.pal(7,"Dark2"),c("SK","AB","BC","MB","ON","QC","NS")),
setNames(RColorBrewer::brewer.pal(3,"Dark2"),c("NL","PE","NB")))
successful_label <- "Atlantic provinces sans NL"
```
```{r eval=FALSE, include=FALSE}
plot_data <- covid_data %>%
select(Date,Province,shortProvince,Cases,update) %>%
complete(Date,Province) %>%
mutate(Cases=replace_na(Cases,0)) %>%
left_join(pop_data,by="Province") %>%
group_by(Province) %>%
arrange(desc(Date)) %>%
filter(cumsum(Cases)>0) %>% # remove trailing zeros
arrange(Date) %>%
mutate(Cases=clean_missing_weekend_data(Cases)) %>%
mutate(incidence=roll::roll_sum(Cases,7)/Population*100000) %>%
mutate(type=ifelse(shortProvince %in% successful_provinces,"Atlantic bubble & Territories","Rest of Canada"))
plot_data %>%
filter(Date>=as.Date("2020-03-01"),shortProvince!="CAN") %>%
#filter(Date>=as.Date("2020-11-01")) %>%
ggplot(aes(x=Date,y=incidence,group=shortProvince)) +
geom_line(data=~filter(.,shortProvince %in% successful_provinces),
color="grey") +
geom_point(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
color="grey") +
ggrepel::geom_text_repel(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
aes(label=shortProvince),nudge_x = 50,min.segment.length = 0,color="grey",direction="y") +
geom_line(data=~filter(.,!(shortProvince %in% successful_provinces)),
aes(color=shortProvince)) +
geom_point(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince)) +
ggrepel::geom_text_repel(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince,label=shortProvince),nudge_x = 50,min.segment.length = 0,direction="y") +
scale_color_brewer(palette = "Dark2",guide=FALSE) +
facet_wrap("type",ncol=1,scales ="free_y") +
expand_limits(x=max(plot_data$Date)+7) +
theme_bw() +
scale_x_date(breaks="months",labels=function(d)strftime(d,"%b")) +
labs(title=paste0("7 day incidence for Canadian provinces (as of ",plot_data$Date %>% last,")"),
x=NULL,y="Cumulative 7 day cases per 100k population",
color=NULL,
caption="MountainMath, Data: PHAC")
```
The Atlantic provinces have pursued very different COVID-19 strategies from the other provinces and have seen very different outcomes. The 7-day incidence, that is the cumulative number of cases over the past 7 days per 100,000 population, has been used by many jurisdictions as a key metric to trigger policy interventions.
```{r two-covid-canadas-overview}
plot_data <- covid_data %>%
select(Date,Province,shortProvince,Cases,TotalDeaths=Deaths,Deaths=numdeathstoday,update) %>%
complete(Date,Province) %>%
mutate(Cases=replace_na(Cases,0),
Deaths=replace_na(Deaths,0)) %>%
left_join(pop_data,by="Province") %>%
group_by(Province) %>%
arrange(desc(Date)) %>%
filter(cumsum(update)>0) %>% # remove trailing zeros
arrange(Date) %>%
mutate(Cases=clean_missing_weekend_data(Cases)) %>%
mutate(incidence=roll::roll_sum(Cases,7)/Population*100000) %>%
mutate(type=ifelse(shortProvince %in% successful_provinces,successful_label,"Other provinces"))
g <- plot_data %>%
filter(Date>=as.Date("2020-03-01"),shortProvince!="CAN") %>%
filter(!(shortProvince %in% c("NT","YT","NU"))) %>%
#filter(Date>=as.Date("2020-11-01")) %>%
ggplot(aes(x=Date,y=incidence,group=shortProvince,color=type)) +
geom_line() +
#geom_point(shape=21) +
scale_color_manual(values=sanzo::duos$c079 %>% rev) +
#facet_wrap("type",ncol=1,scales ="free_y") +
expand_limits(x=max(plot_data$Date)+7) +
theme_bw() +
theme(legend.position="bottom") +
scale_x_date(breaks="months",labels=function(d)strftime(d,"%b")) +
labs(title="Two COVID Canadas",
x=NULL,y="Cumulative 7 day cases per 100k population",
color=NULL,
caption="MountainMath, Data: PHAC")
g
#r<-graph_to_s3(g,"bccovid","two-covid-canadas.png",width=knitr::opts_chunk$get('fig.width'),height=knitr::opts_chunk$get('fig.height'))
```
For better comparison we can plot the Atlantic provinces and the other provinces on different scales.
```{r two-covid-canadas}
plot_data <- covid_data %>%
filter(Province!="Repatriated") %>%
select(Date,Province,shortProvince,Cases,TotalDeaths=Deaths,Deaths=numdeathstoday,update) %>%
# complete(Date,Province) %>%
# mutate(Cases=replace_na(Cases,0),
# Deaths=replace_na(Deaths,0)) %>%
left_join(pop_data,by="Province") %>%
group_by(Province) %>%
arrange(desc(Date)) %>%
filter(cumsum(update)>0) %>% # remove trailing zeros
arrange(Date) %>%
mutate(Cases=clean_missing_weekend_data(Cases)) %>%
mutate(incidence=roll::roll_sum(Cases,7)/Population*100000) %>%
#mutate(incidence=zoo::rollsum(Cases,7,align="center",fill=as.numeric(NA))/Population*100000) %>%
mutate(type=ifelse(shortProvince %in% successful_provinces,successful_label,"Other provinces"))
g <- plot_data %>%
filter(Date>=as.Date("2020-03-01"),shortProvince!="CAN") %>%
filter(!(shortProvince %in% c("NT","YT","NU"))) %>%
#filter(Date>=as.Date("2020-11-01")) %>%
ggplot(aes(x=Date,y=incidence,group=shortProvince)) +
geom_line(data=~filter(.,shortProvince %in% successful_provinces),
aes(color=shortProvince)) +
geom_point(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
aes(color=shortProvince)) +
ggrepel::geom_text_repel(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
aes(label=shortProvince,color=shortProvince),nudge_x = 15,direction="y",
segment.colour="darkgrey") +
geom_line(data=~filter(.,!(shortProvince %in% successful_provinces)),
aes(color=shortProvince)) +
geom_point(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince)) +
ggrepel::geom_text_repel(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince,label=shortProvince),nudge_x = 15,direction="y",
segment.colour="darkgrey") +
#scale_color_brewer(palette = "Dark2",guide=FALSE) +
scale_color_manual(values=province_colours,guide=FALSE) +
facet_wrap("type",ncol=1,scales ="free_y") +
expand_limits(x=max(plot_data$Date)+7) +
theme_bw() +
scale_x_date(breaks="months",labels=function(d)strftime(d,"%b")) +
labs(title="Two COVID Canadas",
x=NULL,y="Cumulative 7 day cases per 100k population",
color=NULL,
caption="MountainMath, Data: PHAC")
g
#r<-graph_to_s3(g,"bccovid","two-covid-canadas-overview.png",width=knitr::opts_chunk$get('fig.width'),height=knitr::opts_chunk$get('fig.height'))
```
## Trend lines
Sometimes it is useful to get a clearer view on trend lines. Rolling 7-day sums (like above) or rolling averages (as often emplyed) are a problematic way to represent trend lines as the lag actual trends by 3 days.
A fairly simple trend line model like a (multiplicative) STL decomposition can extract cleaner trend lines that also cover the most recent 3 days of data, at the expense of a bit of added volatility at the very end of the trend line where the trend line may shift slightly when new data comes in.
```{r two-covid-canadas-trend}
successful_provinces <- c("NL","NT","NS","YT","PE","NB")
province_colours <- c(setNames(RColorBrewer::brewer.pal(6,"Dark2"),c("SK","AB","BC","MB","ON","QC")),
setNames(RColorBrewer::brewer.pal(4,"Dark2"),c("NL","NS","PE","NB")))
successful_label <- "Atlantic provinces"
plot_data <- covid_data %>%
filter(Province!="Repatriated") %>%
select(Date,Province,shortProvince,Cases,TotalDeaths=Deaths,Deaths=numdeathstoday,update) %>%
#complete(Date,Province) %>%
mutate(Cases=replace_na(Cases,0),
Deaths=replace_na(Deaths,0)) %>%
left_join(pop_data,by="Province") %>%
group_by(Province) %>%
arrange(desc(Date)) %>%
filter(cumsum(update)>0) %>% # remove trailing zeros
arrange(Date) %>%
mutate(Cases=clean_missing_weekend_data(Cases)) %>%
mutate(Cases=pmax(0,Cases)) %>%
mutate(trend=(extract_stl_trend_m(Cases+1)-1)/Population*100000) %>%
mutate(type=ifelse(shortProvince %in% successful_provinces,successful_label,"Other provinces"))
g <- plot_data %>%
filter(Date>=as.Date("2020-03-01"),shortProvince!="CAN") %>%
filter(!(shortProvince %in% c("NT","YT","NU"))) %>%
#filter(Date>=as.Date("2020-11-01")) %>%
ggplot(aes(x=Date,y=trend,group=shortProvince)) +
geom_line(data=~filter(.,shortProvince %in% successful_provinces),
aes(color=shortProvince)) +
geom_point(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
aes(color=shortProvince)) +
ggrepel::geom_text_repel(data=~filter(.,shortProvince %in% successful_provinces,Date==max(Date)),
aes(label=shortProvince,color=shortProvince),nudge_x = 15,direction="y",
segment.colour="darkgrey") +
geom_line(data=~filter(.,!(shortProvince %in% successful_provinces)),
aes(color=shortProvince)) +
geom_point(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince)) +
ggrepel::geom_text_repel(data=~filter(.,!(shortProvince %in% successful_provinces),Date==max(Date)),
aes(color=shortProvince,label=shortProvince),nudge_x = 15,direction="y",
segment.colour="darkgrey") +
#scale_color_brewer(palette = "Dark2",guide=FALSE) +
scale_color_manual(values=province_colours,guide=FALSE) +
facet_wrap("type",ncol=1,scales ="free_y") +
expand_limits(x=max(plot_data$Date)+7) +
theme_bw() +
scale_x_date(breaks="months",labels=function(d)strftime(d,"%b")) +
labs(title="Two COVID Canadas (STL trend lines)",
x=NULL,y="Daily cases per 100k population",
color=NULL,
caption="MountainMath, Data: PHAC")
g
#r<-graph_to_s3(g,"bccovid","two-covid-canadas-overview.png",width=knitr::opts_chunk$get('fig.width'),height=knitr::opts_chunk$get('fig.height'))
```