README.qmd

---
title: Modelling traffic in Capetown
format: gfm
execute: 
  cache: true
  echo: false
  message: false
  warning: false
---

```{r}
#| label: setup
#| include: false
# Set cran mirror:
options(repos = c(CRAN = "https://cloud.r-project.org"))
if (!require("remotes")) install.packages("remotes")
pkgs = c(
    "sf",
    "tidyverse",
    "osmextract",
    "zonebuilder",
    "dodgr",
    "tmap",
    "webshot2",
    "geodist"
)
remotes::install_cran(pkgs)
sapply(pkgs, require, character.only = TRUE)
tmap_mode("view")
```

```{r}
#| label: case-study-area
# case_study_area = zonebuilder::zb_zone("Cape Town", n_circles = 4) # 10 km radius
case_study_area = zonebuilder::zb_zone("Cape Town", n_circles = 6) # 21 km radius
```

```{r}
m = qtm(case_study_area)
```

```{r}
```
```{r}
#| eval: false
streetnet = dodgr::dodgr_streetnet(
    sf::st_bbox(case_study_area)
)
```

```{r}
# To overcome this issue:
# Error: Overpass query unavailable without internet
assign("has_internet_via_proxy", TRUE, environment(curl::has_internet))
```

```{r}
if (!file.exists("streetnet.rds")) {
    streetnet = dodgr::dodgr_streetnet(
        sf::st_bbox(case_study_area)
    )
    saveRDS(streetnet, "streetnet.rds")
}
streetnet = readRDS("streetnet.rds")
```


```{r}
#| label: convert-to-graph
# names(streetnet) many OSM tags
# graph = dodgr::weight_streetnet(streetnet)
cols_to_keep = c(
    "osm_id",
    "highway",
    "oneway",
    "maxspeed",
    "lanes",
    "lit"
)
graph = dodgr::weight_streetnet(streetnet, wt_profile = "motorcar", keep_cols = cols_to_keep)
```


```{r}
#| label: post-processing
clear_dodgr_cache()
graph = dodgr::dodgr_deduplicate_graph(graph)
# (dodgr:::duplicated_edge_check (graph))
graph_contracted = dodgr::dodgr_contract_graph(graph)
dodgr:::duplicated_edge_check(graph_contracted)
nrow(graph_contracted) / nrow(graph) # 1/3rd size
```


```{r}
#| label: centrality-calc
graph_centrality = dodgr::dodgr_centrality(graph_contracted)
```


```{r}
graph_sf = dodgr::dodgr_to_sf(graph)
```


```{r}
#| label: prepare-df-centrality
names(graph_centrality)
df_centrality =  tibble(
    edge_id = graph_centrality$edge_id,
    centrality = graph_centrality$centrality
)
head(df_centrality)
```


```{r}
graph_joined = left_join(graph_sf, df_centrality)
```


```{r}
#| label: plot-centrality
case_study_area_1km = case_study_area[1, ]
graph_joined_1km = graph_joined[case_study_area_1km, op = sf::st_within]
graph_joined_1km |>
  ggplot() +
  geom_sf(aes(colour = centrality)) +
  scale_colour_viridis_c(transform = "log") +
  theme_void()
```

Let's plot the road segments with the top 1% of centrality values:

```{r}
#| label: plot-centrality-top-30
graph_joined |>
#   filter(centrality > quantile(centrality, 0.70, na.rm = TRUE)) |>
  ggplot() +
  geom_sf(aes(alpha = log(centrality))) +
  theme_void()
```

```{r}
#| label: centrality-distribution
graph_joined |>
  ggplot() +
  geom_histogram(aes(centrality)) +
  scale_x_log10() 
```


The dataset with the centrality added is as follows:

```{r}
graph_joined |>
  sf::st_drop_geometry() |>
  head() |>
  knitr::kable()
```

We can proceed to generate a model of traffic based on the centrality of the roads if we have a training dataset with traffic data.