coronavirus - dataviz avec R

Bonjour.

Je n'ai pas les compétences pour faire un modèle de prédiction de la fin du confinement.

:fleche: Ce qui suis n'a pas l'objectif de prédire ni de mener a une quelconque conclusion juste a me distraire de mon insomnie. Et pourquoi pas d'inviter à faire vos propres visualisations de donnée, et éventuellement les partager.

Le lien suivant est donné dans un des PDF fournis par gerard093:

https://opendata.ecdc.europa.eu/covi...stribution/csv

0/ Modules:

Code:

1
2
3
4
5
 
library(readr)
library(RSQLite)
library(ggplot2)
library(gridExtra)

1/ Collecter les data:

Code:

1
2
3
4
5
6
7
8
9
10
11
12
 
# collecte data brute
d <- read_csv("https://opendata.ecdc.europa.eu/covid19/casedistribution/csv", 
                     col_types = cols(dateRep = col_date(format = "%d/%m/%Y"))
              )
 
# typage
d$dateRep <- as.factor(d$dateRep)
d$countriesAndTerritories <- as.factor(d$countriesAndTerritories)
d$geoId <- as.factor(d$geoId)
d$countryterritoryCode <- as.factor(d$countryterritoryCode)
d$continentExp <- as.factor(d$continentExp)

2/ Comme je ne suis pas encore alèse avec dplyr je passe par du SQLite via R

Code:

1
2
3
4
5
6
 
# ouverture connexion sur une SQLite en memoire vive
c <- dbConnect(RSQLite::SQLite(), dbname = ":memory:")
 
# ecriture des data brute dans un table "data"
dbWriteTable(c, name = "data", value = d)

3/ Confection d'une data.frame qui cumul ( jour par jour ) le nombre de cas (Covid-19) par continent:

Code:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
 
# creation d'une requete pour cumuler les "cases" et "deaths" par continent
t <- dbSendQuery(c, "SELECT date(dateRep) AS dateRep
                          , continentExp  AS continent
                          , SUM(cases) OVER ( PARTITION BY continentExp 
                                              ORDER BY date(dateRep)) AS cum_c
                          , SUM(deaths) OVER ( PARTITION BY continentExp 
                                              ORDER BY date(dateRep)) AS cum_d
                     FROM data")
 
# recuperation du resultat de la requete dans un data.frame
tt <- dbFetch(t)
 
# liberation de la requete
dbClearResult(t)
 
# typage date
tt$dateRep <- as.POSIXct(tt$dateRep)

4/ Graphique:

Code:

1
2
3
4
5
6
7
8
9
10
11
12
13
 
# definition du graph
ggplot(tt) +
  aes(x = tt$dateRep, ) +
  geom_line(aes(y = tt$cum_d), col = "red") +
  geom_line(aes(y = tt$cum_c), col = "blue") +
  facet_wrap( tt$continent ~ . , scales = "free_y") +
  labs(
    x = NULL,
    y = "Nb",
    title = "Nombre de cas / mort cumulé par continent",
    subtitle = "Les cas en Bleu / Les morts en Rouge"
  )

Pièce jointe 560914

5/ Autre graphique:

Code:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
 
# nombre de cas / continent
g1 <- ggplot(tt) +
  aes(x = dateRep, y = cum_c, colour = continent  ) +
  geom_line() +
  labs(
    x = NULL,
    y = "Nombre (cumulé)",
    title = "Cas"
  )
 
# nombre de morts / continent
g2 <- ggplot(tt) +
  aes(x = dateRep, y = cum_d, colour = continent ) +
  geom_line() +
  labs(
    x = NULL,
    y = NULL,
    title = "Morts"
  )
 
g12 <- grid.arrange(g1, g2, ncol = 2, nrow = 1)

Pièce jointe 560918

6/ Focus sur la France:

Code:

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
 
### Focus sur la france
f <- dbSendQuery(c, "SELECT date(dateRep) AS dateRep
                          , countriesAndTerritories AS pays
                          , SUM(cases) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_c
                          , SUM(deaths) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_d
                     FROM data
                     WHERE countriesAndTerritories = 'France'")
 
ff <- dbFetch(f)
dbClearResult(f)
ff$dateRep <- as.POSIXct(ff$dateRep)
ggplot(ff) +
  aes(x = dateRep, ) +
  geom_line(aes(y = cum_d), col = "red") +
  geom_line(aes(y = cum_c), col = "blue") +
  labs(
    x = NULL,
    y = "Nb",
    title = "Nombre de cas / mort cumulé pour la France",
    subtitle = "Les cas en Bleu / Les morts en Rouge"
  ) +
  geom_rug(aes(y = cum_d), col = "red", sides = "l") +
  geom_rug(aes(y = cum_c), col = "blue",sides = "l")

Pièce jointe 560920

7/ "Top" 7 des pays (en cas et mort) du continent Europe:

Code:

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
 
### Focus sur l'europe Top 7
e <- dbSendQuery(c, "WITH d AS (SELECT date(dateRep) AS dateRep
                          , countriesAndTerritories AS pays
                          , SUM(cases) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_c
                          , SUM(deaths) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_d
                     FROM data
                     WHERE continentExp = 'Europe'
                     )
                     SELECT *
                     FROM d
                     WHERE pays IN (
                       SELECT pays 
                       FROM (
                         SELECT pays
                              , MAX(cum_d)AS n
                         FROM d
                         GROUP BY pays
                         ORDER BY n DESC
                         LIMIT 7
                       )
                     )
                     ")
 
e1 <- dbFetch(e)
dbClearResult(e)
e1$dateRep <- as.POSIXct(e1$dateRep)
ge1 <- ggplot(e1) +
  aes(x = dateRep, y = cum_d, color = pays ) +
  geom_line() +
  labs(
    x = NULL,
    y = NULL,
    title = "Nombre de morts"
  )
 
###
e <- dbSendQuery(c, "WITH d AS (SELECT date(dateRep) AS dateRep
                          , countriesAndTerritories AS pays
                          , SUM(cases) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_c
                          , SUM(deaths) OVER ( PARTITION BY countriesAndTerritories
                                              ORDER BY date(dateRep)) AS cum_d
                     FROM data
                     WHERE continentExp = 'Europe'
                     )
                     SELECT *
                     FROM d
                     WHERE pays IN (
                       SELECT pays 
                       FROM (
                         SELECT pays
                              , MAX(cum_c)AS n
                         FROM d
                         GROUP BY pays
                         ORDER BY n DESC
                         LIMIT 7
                       )
                     )
                     ")
 
e2 <- dbFetch(e)
dbClearResult(e)
e2$dateRep <- as.POSIXct(e2$dateRep)
ge2 <- ggplot(e2) +
  aes(x = dateRep, y = cum_c, color = pays ) +
  geom_line() +
  labs(
    x = NULL,
    y = NULL,
    title = "Nombre de cas"
  )
 
e12 <- grid.arrange(ge1, ge2, ncol = 1, nrow = 2)

Pièce jointe 560931

7.1/ Pourcentage de mort en fonction du nombre de cas (cumulé).

Code:

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
 
e <- dbSendQuery(c, "WITH d AS (SELECT dateRep,
                            pays,
                            A.cum_c / pop_2018 * 100 AS per_c,
                            A.cum_d / A.cum_c * 100 AS per_m
                     FROM (
                       SELECT date(dateRep) AS dateRep
                            , countriesAndTerritories AS pays
                            , SUM(cases) OVER ( PARTITION BY countriesAndTerritories
                                                ORDER BY date(dateRep)) AS cum_c
                            , SUM(deaths) OVER ( PARTITION BY countriesAndTerritories
                                                ORDER BY date(dateRep)) AS cum_d
                            , popData2018 AS pop_2018
                       FROM data
                       WHERE continentExp = 'Europe'
                     ) AS A
                    )
 
                     SELECT *
                     FROM d
                     WHERE pays IN (
                       SELECT pays 
                       FROM (
                         SELECT pays
                              , MAX(per_m) AS n
                         FROM d
                         GROUP BY pays
                         ORDER BY n DESC
                         LIMIT 7
                       )
                     )
 
                     ")
 
ee <- dbFetch(e)
dbClearResult(e)
ee
ee$dateRep <- as.POSIXct(ee$dateRep)
 
ggplot(ee) +
  aes(x = dateRep, y = per_m, color = pays) +
  geom_line() +
  labs(
    x = NULL,
    y = "%",
    title = "Pourcentage de mort en fonction du nombre de cas (cumulé)"
  )

Pièce jointe 560936

7.2/ Pourcentage de cas cumulé en fonction de la population du pays:

Code:

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
 
e <- dbSendQuery(c, "WITH d AS (SELECT dateRep,
                            pays,
                            A.cum_c / pop_2018 * 100 AS per_c,
                            A.cum_d / A.cum_c * 100 AS per_m
                     FROM (
                       SELECT date(dateRep) AS dateRep
                            , countriesAndTerritories AS pays
                            , SUM(cases) OVER ( PARTITION BY countriesAndTerritories
                                                ORDER BY date(dateRep)) AS cum_c
                            , SUM(deaths) OVER ( PARTITION BY countriesAndTerritories
                                                ORDER BY date(dateRep)) AS cum_d
                            , popData2018 AS pop_2018
                       FROM data
                       WHERE continentExp = 'Europe'
                     ) AS A
                    )
 
                     SELECT *
                     FROM d
                     WHERE pays IN (
                       SELECT pays 
                       FROM (
                         SELECT pays
                              , MAX(per_c) AS n
                         FROM d
                         GROUP BY pays
                         ORDER BY n DESC
                         LIMIT 7
                       )
                     )
 
                     ")
 
ee <- dbFetch(e)
dbClearResult(e)
ee
ee$dateRep <- as.POSIXct(ee$dateRep)
ggplot(ee) +
  aes(x = dateRep, y = per_c, color = pays) +
  geom_line() +
  labs(
    x = NULL,
    y = "%",
    title = "Pourcentage de cas cumulé en fonction de la population du pays"
  )

Pièce jointe 560927