Enhance tp2 Rmd (histogram, interactive maps, choropleth fixes); add TP3 Shiny apps and project files; update .gitignore; add shap (+cloudpickle, numba, llvmlite, tqdm, slicer) to pyproject.toml and uv.lock; remove generated tp2 HTML/assets

M2/Data Visualisation/tp2/tp2.Rmd

@@ -48,7 +48,6 @@ library(tidyr)
 library(rmarkdown)
 library(ggthemes)
 library(cowplot)
-
 ```
 
 # Objectifs du TP
@@ -91,7 +90,7 @@ ont lieu tous les 5 ans.
 Dans un premier temps, il faut installer le package et le charger.
 
 ```{r}
-# install.packages("gapminder") #nolint
+# install.packages("gapminder")
 library(gapminder)
 ```
 
@@ -159,6 +158,12 @@ ggplot(data = gapminder, aes(x = lifeExp)) +
   geom_density()
 ```
 
+```{r}
+ggplot(data = gapminder, aes(x = lifeExp)) +
+  geom_histogram(aes(y = ..density..), bins = 30) +
+  geom_density()
+```
+
 ### Cartes
 
 Afin de visualiser plus précisément les caractéristiques régionales de
@@ -226,8 +231,9 @@ library(tidyverse)
 path <- getwd()
 accidents <- read_csv(
   paste0(
-         path,
-         "/data/accidentsVelo.csv"),
+    path,
+    "/data/accidentsVelo.csv"
+  ),
   col_types = cols(Num_Acc = col_double(), date = col_date(format = "%Y-%m-%d"))
 )
 
@@ -245,8 +251,13 @@ accidents <- accidents |>
 # correct some issues with variables `hrmn`
 issue <- which(str_length(accidents$hrmn) == 4)
 correct <- accidents$hrmn[issue]
-correct <- paste0("0", str_sub(correct, 1, 1), ":",
-                  str_sub(correct, 2, 2), str_sub(correct, 4, 4))
+correct <- paste0(
+  "0",
+  str_sub(correct, 1, 1),
+  ":",
+  str_sub(correct, 2, 2),
+  str_sub(correct, 4, 4)
+)
 accidents$hrmn[issue] <- correct
 
 # Extract hour
@@ -257,8 +268,9 @@ accidents <- accidents |>
 # mapping table for french departments
 departements_francais <- read_excel(
   paste0(
-         path,
-         "/data/departements-francais.xlsx"),
+    path,
+    "/data/departements-francais.xlsx"
+  ),
   col_types = c("text", "text", "text")
 )
 ```
@@ -321,9 +333,57 @@ intéressantes sont à extraire de ce jeu de données.
     (latitude, longitude) en France sur toute la période en modulant la
     gravité.
 
+```{r}
+library(mapview)
+library(sf)
+
+## Remove NA
+df_map_dyn <- accidents |>
+  filter(!is.na(lat) & !is.na(long)) |>
+  mutate(
+    lat = as.numeric(str_replace_all(str_trim(lat), ",", ".")),
+    long = as.numeric(str_replace_all(str_trim(long), ",", "."))
+  ) |>
+  filter(!is.na(lat) & !is.na(long))
+
+# Make map and print it
+mymap <- st_as_sf(df_map_dyn[1:5000, ], coords = c("long", "lat"), crs = 4326)
+mapview(
+  mymap,
+  cex = 2,
+  legend = TRUE,
+  layer.name = "Gravité",
+  zcol = "grav",
+  map.types = "OpenStreetMap"
+)
+```
+
 2.  Faire deux variantes de cette carte, selon le caractère urbain ou
     non des accidents.
 
+```{r}
+# Map for urban accidents
+df_map_urban <- df_map_dyn |>
+  mutate(
+    agg = dplyr::if_else(
+      agg == 1,
+      "urbanisation",
+      "agglomération",
+      missing = NA_character_
+    )
+  )
+
+# Make map and print it
+mymap_urban <- st_as_sf(df_map_urban, coords = c("long", "lat"), crs = 4326)
+mapview(
+  mymap_urban,
+  cex = 2,
+  legend = TRUE,
+  layer.name = "Urbanisation",
+  zcol = "agg"
+)
+```
+
 3.  Commenter ces figures.
 
 4.  Quelles limites voyez-vous à cette représentation ?
@@ -333,21 +393,6 @@ Voici un premier code à trou pour vous aider. Pour alléger les temps de
 production afficher uniquement quelques points. Vous pourrez ajouter
 l'ensemble du jeu de données quand votre code sera finalisé.
 
-```{r}
-library(mapview)
-library(sf)
-
-## Remove NA
-df_map_dyn <- accidents |>
-  filter(!is.na(lat) & !is.na(long)) |>
-  na.omit()
-
-# Make map and print it
-mymap <- st_as_sf(df_map_dyn, coords=c("long", "lat"), crs=4326)
-mapview(mymap, cex = 2, layer.name = "Gravité",
-        zcol = "grav",legend = TRUE )
-```
-
 #### Carte choroplèthe {.unnumbered}
 
 ::: exercise-box
@@ -360,9 +405,9 @@ mapview(mymap, cex = 2, layer.name = "Gravité",
 
 Voici un premier code à trou pour vous aider.
 
-```{r, eval = F}
-# get french map - level nuts2
-fr <- gisco_get_nuts(resolution = "20", country = ???, nuts_level = ???) |>
+```{r}
+# get french map - level nuts3
+fr <- gisco_get_nuts(resolution = "20", country = "FRA", nuts_level = 3) |>
   mutate(res = "20M")
 
 # Remove white-space to avoid errors.
@@ -375,23 +420,28 @@ fr <- fr |>
 
 # Merge and remove departements outside metropolitan France
 fr_map <- fr |>
-  left_join(???) |>
-  filter(! dep %in% c("971", ???) )
+  left_join(departements_francais, by = c("NUTS_NAME" = "dep_name")) |>
+  filter(!dep %in% c("971", "972", "973", "974", "976"))
 
 # count the number of accidents
-df_acc <-  ???
+df_acc <- accidents |>
+  filter(!is.na(dep)) |>
+  group_by(dep) |>
+  summarise(n = n())
 
 # merge statistics with the map
 map_acc <- fr_map |>
-  left_join(df_acc, by = c("dep" =  "dep"))
+  left_join(df_acc, by = c("dep" = "dep"))
 
 # map with all accidents
 g_map_acc <- ggplot(map_acc) +
-  geom_sf(???) +
-   scale_fill_viridis_c(option = "viridis") +
-  labs(title = "Carte des accidents de vélo",
-       subtitle = "Année 2005-2021",
-       fill = "Nombre d'accidents") +
+  geom_sf(aes(fill = n)) +
+  scale_fill_viridis_c(option = "viridis") +
+  labs(
+    title = "Carte des accidents de vélo",
+    subtitle = "Année 2005-2021",
+    fill = "Nombre d'accidents"
+  ) +
   theme_void()
 g_map_acc
 ```