Pour le #Tidytuesday de cette semaine, on regarde les statistiques de la communauté R4DS avec des graphiques en escalier. Au menu: tendance, corrélation et mini-dashboard pour visualiser les données.
CONTEXTE
Le R4DS est une communauté en ligne pour toutes les personnes qui travaillent avec R et qui désirent améliorer leur compétences. Plus tôt ce mois-ci, la fondatrice, Jesse Mostipak, a donnée une présentation à la conférence de useR-2019 de Toulouse (France). On peut rejoindre cette communauté sur Slack.
OBJECTIFS
1) Explorer les données avec inspectdf
2) Visualiser l’évolution du nombre de membres depuis la création de la communauté avec un graphique en escaliser
3) Visualiser l’activité des membres depuis la création de la communauté aussi avec un graphique en escalier
4) Créer un mini-dashboard avec mes deux graphiques pour créer une viz qui explique mon message
IMPORTER
r4ds_members <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-07-16/r4ds_members.csv")FALSE Parsed with column specification:
FALSE cols(
FALSE .default = col_double(),
FALSE date = col_date(format = "")
FALSE )FALSE See spec(...) for full column specifications.
EXPLORER
glimpse(r4ds_members)## Observations: 678
## Variables: 21
## $ date <date> 2017-08-27, 2017-08-28, 20…
## $ total_membership <dbl> 1, 1, 1, 1, 1, 188, 284, 32…
## $ full_members <dbl> 1, 1, 1, 1, 1, 188, 284, 32…
## $ guests <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ daily_active_members <dbl> 1, 1, 1, 1, 1, 169, 225, 21…
## $ daily_members_posting_messages <dbl> 1, 0, 1, 0, 1, 111, 110, 96…
## $ weekly_active_members <dbl> 1, 1, 1, 1, 1, 169, 270, 30…
## $ weekly_members_posting_messages <dbl> 1, 1, 1, 1, 1, 111, 183, 21…
## $ messages_in_public_channels <dbl> 4, 0, 0, 0, 1, 252, 326, 20…
## $ messages_in_private_channels <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ messages_in_shared_channels <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ messages_in_d_ms <dbl> 1, 0, 0, 0, 0, 119, 46, 71,…
## $ percent_of_messages_public_channels <dbl> 0.8000, 0.0000, 0.0000, 0.0…
## $ percent_of_messages_private_channels <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ percent_of_messages_d_ms <dbl> 0.2000, 0.0000, 0.0000, 0.0…
## $ percent_of_views_public_channels <dbl> 0.2857, 1.0000, 1.0000, 1.0…
## $ percent_of_views_private_channels <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ percent_of_views_d_ms <dbl> 0.7143, 0.0000, 0.0000, 0.0…
## $ name <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ public_channels_single_workspace <dbl> 10, 10, 11, 11, 12, 12, 12,…
## $ messages_posted <dbl> 35, 35, 37, 38, 66, 1101, 1…summary(r4ds_members)## date total_membership full_members guests
## Min. :2017-08-27 Min. : 1.0 Min. : 1.0 Min. :0
## 1st Qu.:2018-02-12 1st Qu.: 978.2 1st Qu.: 978.2 1st Qu.:0
## Median :2018-07-31 Median :1605.0 Median :1605.0 Median :0
## Mean :2018-07-31 Mean :1567.8 Mean :1567.8 Mean :0
## 3rd Qu.:2019-01-16 3rd Qu.:2142.8 3rd Qu.:2142.8 3rd Qu.:0
## Max. :2019-07-05 Max. :3029.0 Max. :3029.0 Max. :0
## daily_active_members daily_members_posting_messages weekly_active_members
## Min. : 1.00 Min. : 0.00 Min. : 1.0
## 1st Qu.: 63.00 1st Qu.: 6.00 1st Qu.:206.0
## Median : 88.00 Median : 11.00 Median :239.0
## Mean : 91.39 Mean : 13.24 Mean :249.7
## 3rd Qu.:110.00 3rd Qu.: 16.00 3rd Qu.:307.8
## Max. :258.00 Max. :111.00 Max. :525.0
## weekly_members_posting_messages messages_in_public_channels
## Min. : 1.00 Min. : 0.00
## 1st Qu.: 35.00 1st Qu.: 9.25
## Median : 48.00 Median : 19.00
## Mean : 52.16 Mean : 28.46
## 3rd Qu.: 59.00 3rd Qu.: 35.00
## Max. :278.00 Max. :326.00
## messages_in_private_channels messages_in_shared_channels messages_in_d_ms
## Min. : 0.000 Min. :0 Min. : 0.00
## 1st Qu.: 0.000 1st Qu.:0 1st Qu.: 1.00
## Median : 0.000 Median :0 Median : 4.00
## Mean : 1.718 Mean :0 Mean : 13.05
## 3rd Qu.: 0.000 3rd Qu.:0 3rd Qu.: 12.00
## Max. :75.000 Max. :0 Max. :227.00
## percent_of_messages_public_channels percent_of_messages_private_channels
## Min. :0.0000 Min. :0.0000
## 1st Qu.:0.5840 1st Qu.:0.0000
## Median :0.8000 Median :0.0000
## Mean :0.7248 Mean :0.0305
## 3rd Qu.:0.9444 3rd Qu.:0.0000
## Max. :1.0000 Max. :1.0000
## percent_of_messages_d_ms percent_of_views_public_channels
## Min. :0.0000 Min. :0.2726
## 1st Qu.:0.0345 1st Qu.:0.9115
## Median :0.1595 Median :0.9519
## Mean :0.2270 Mean :0.9285
## 3rd Qu.:0.3478 3rd Qu.:0.9744
## Max. :1.0000 Max. :1.0000
## percent_of_views_private_channels percent_of_views_d_ms name
## Min. :0.000000 Min. :0.00000 Min. :0
## 1st Qu.:0.000000 1st Qu.:0.02235 1st Qu.:0
## Median :0.000000 Median :0.04170 Median :0
## Mean :0.009773 Mean :0.06176 Mean :0
## 3rd Qu.:0.006450 3rd Qu.:0.07433 3rd Qu.:0
## Max. :0.267400 Max. :0.72170 Max. :0
## public_channels_single_workspace messages_posted
## Min. :10.0 Min. : 35
## 1st Qu.:15.0 1st Qu.:20543
## Median :19.0 Median :33828
## Mean :17.8 Mean :32936
## 3rd Qu.:21.0 3rd Qu.:40104
## Max. :27.0 Max. :59627Pas de données manquante ici dans les données et les variables sont stockés sous le bon format. Pour visualiser rapidement les corrélations entre les variables, j’ai dû réduire le nombre de variables et utiliser seulement celles qui me semblaient le plus pertinentes.
r4ds_cor<-r4ds_members %>%
select('total_membership', 'daily_active_members', 'daily_members_posting_messages', 'weekly_active_members', 'weekly_members_posting_messages', 'messages_posted') %>%
inspect_cor() %>%
show_plot()
Comme on pouvait s’y attendre la corrélation est très forte entre le nombre de message publié et le nombre de membre…
PRÉPARER:
r4ds<-r4ds_members %>%
select('date','total_membership','messages_posted') %>%
mutate(quarter=quarter(date, with_year = TRUE)) %>%
group_by(quarter) %>%
summarise(tot_m_1000=sum(total_membership/1000)) %>%
filter(!quarter %in% 2019.3)
r4ds_point<-r4ds %>%
filter(quarter %in% c(2017.3, 2019.2))
r4ds_active<-r4ds_members %>%
select('date','total_membership','daily_active_members') %>%
mutate(quarter=quarter(date, with_year = TRUE)) %>%
group_by(quarter) %>%
summarise(active_1000=sum(daily_active_members/1000)) %>%
filter(!quarter %in% 2019.3)
r4ds_point_active<-r4ds_active %>%
filter(quarter %in% c(2017.3, 2019.2))
VISUALISER
#Graphique
gg1<-ggplot(data=r4ds, aes(x = quarter, y=tot_m_1000))
gg1<-gg1 + geom_step(linetype=5, color="#A9A9A9", size=2.5)
gg1<-gg1 + geom_step(data=r4ds_active, aes(x = quarter, y=active_1000),linetype=5, color="#A9A9A9", size=2.5)
gg1<-gg1 + geom_rect(data=r4ds,
mapping=aes(xmin=2018.1,xmax=2018.4,ymin=0,ymax=Inf),
fill='#01A7C2',alpha=0.05)
gg1<-gg1 + geom_point(data=r4ds_point,
mapping=(aes(x=quarter,y=tot_m_1000)),
color="#A9A9A9", size=5)
gg1<-gg1 + geom_point(data=r4ds_point_active,
mapping=(aes(x=quarter,y=active_1000)),
color="#A9A9A9", size=5)
#ajuster les axes
gg1<-gg1 + scale_x_yearqtr(breaks = seq(from = min(r4ds$quarter), to = max(r4ds$quarter), by = 0.25),
format = "%Y-%q")
gg1<-gg1 + scale_y_continuous(breaks=seq(0,300,50), limits = c(0, 300))
#modifier la légende
gg1<-gg1 + theme(legend.position="none")
#modifier le thème
gg1<-gg1 +theme(panel.border = element_blank(),
panel.background = element_blank(),
plot.background = element_blank(),
panel.grid.major.y= element_blank(),
panel.grid.major.x= element_blank(),
panel.grid.minor = element_blank(),
axis.line.x = element_line(color="#A9A9A9"),
axis.line.y = element_line(color="#A9A9A9"),
axis.ticks= element_blank())
#ajouter les titres
gg1<-gg1 + labs(title="",
subtitle=" ",
y="Membres (x1000)",
x=" ")
gg1<-gg1 + theme(plot.title= element_text(hjust=0,size=15, color="#A9A9A9", face="bold"),
plot.subtitle = element_text(hjust=0,size=12, color="#A9A9A9"),
axis.title.y = element_text(hjust=1,size=12, color="#A9A9A9", angle=90),
axis.title.x = element_blank(),
axis.text.y = element_text(hjust=0.5, size=10, color="#A9A9A9"),
axis.text.x = element_text(hjust=0.5, size=10, color="#A9A9A9"))
#ajouter les étiquettes
gg1<-gg1 + annotate(geom="text", x=2019.2,y=270, label="Total", color="#A9A9A9", size=5, hjust=0.5,vjust=0, fontface="bold")
gg1<-gg1 + annotate(geom="text", x=2019.2,y=18, label="Actif", color="#A9A9A9", size=5, hjust=0.5,vjust=0, fontface="bold")
PRÉPARER:
r4ds<-r4ds_members %>%
mutate(daily_message=messages_posted-shift(messages_posted)) %>%
filter(daily_message>0 & daily_message<5000) %>%
mutate(activity=(daily_message/daily_active_members)) %>%
mutate(quarter=quarter(date, with_year = TRUE)) %>%
group_by(quarter) %>%
summarise(active=sum(activity)) %>%
filter(!quarter %in% 2019.3)
r4ds_point_activity<-r4ds %>%
filter(quarter %in% c(2017.3, 2019.2))
VISUALISER
#Graphique
gg2<-ggplot(data=r4ds, aes(x = quarter, y=active))
gg2<-gg2 + geom_step(linetype=5, color="#A9A9A9", size=2.5)
gg2<-gg2 + geom_rect(data=r4ds,
mapping=aes(xmin=2018.1,xmax=2018.4,ymin=0,ymax=Inf),
fill='#01A7C2',alpha=0.05)
gg2<-gg2 + geom_point(data=r4ds_point_activity,
mapping=(aes(x=quarter,y=active)), color="#A9A9A9", size=5)
#ajuster les axes
gg2<-gg2 + scale_x_yearqtr(breaks = seq(from = min(r4ds$quarter), to = max(r4ds$quarter), by = 0.25),
format = "%Y-%q")
gg2<-gg2 + scale_y_continuous(breaks=seq(0,100,25), limits = c(0, 100))
#modifier la légende
gg2<-gg2 + theme(legend.position="none")
#modifier le thème gg2<-gg2 +theme(panel.border = element_blank(),
panel.background = element_blank(),
plot.background = element_blank(),
panel.grid.major.y= element_blank(),
panel.grid.major.x= element_blank(),
panel.grid.minor = element_blank(),
axis.line.x = element_line(color="#A9A9A9"), axis.line.y = element_line(color="#A9A9A9"),
axis.ticks= element_blank())
#ajouter les titres
gg2<-gg2 + labs(title="",
subtitle=" ",
y="Messages quotidien/membre actif",
x=" ")
gg2<-gg2 + theme(plot.title = element_text(hjust=0,size=15, color="#A9A9A9", face="bold"),
plot.subtitle = element_text(hjust=0,size=12, color="#A9A9A9"),
axis.title.y = element_text(hjust=1,size=12, color="#A9A9A9", angle=90),
axis.title.x = element_blank(),
axis.text.y = element_text(hjust=0.5, size=10, color="#A9A9A9"),
axis.text.x = element_text(hjust=0.5, size=10, color="#A9A9A9"))
Voici ce que ça donne:
Tu veux en savoir plus sur ma démarche? Va écouter l’épisode de podcast dans lequel j’explique mes réflexions pour arriver à ce résultat.