Merge pull request #22 from acteng/21-rewrite-using-od2net-package

Update README to use od2net R package, close #21

.gitignore

@@ -64,3 +64,5 @@ docs
 intermediate/
 *.pbf
 *.tif
+output/output.geojson
+output/*.pmtiles

README.md

@@ -8,9 +8,31 @@ use in the reproducible code below.
 
 # Setup
 
+To run the code below you need to have R and Docker installed. After you
+have installed R, you can install the packages we’ll use as follows:
+
 ``` r
-# # Install geopandas:
-# reticulate::py_install("geopandas")
+# Install pak if not already installed:
+if (!requireNamespace("pak", quietly = TRUE)) {
+  install.packages("pak")
+}
+pkgs = c("sf", "tidyverse", "tmap", "pak")
+pkgs_to_install = pkgs[!pkgs %in% installed.packages()]
+if (length(pkgs_to_install) > 0) {
+  pak::pkg_install(pkgs_to_install)
+}
+# Load the packages with vapply:
+vapply(pkgs, require, logical(1), character.only = TRUE)
+```
+
+           sf tidyverse      tmap       pak 
+         TRUE      TRUE      TRUE      TRUE 
+
+We will also install a couple of package that are not on CRAN:
+
+``` r
+pak::pkg_install("acteng/netgen")
+pak::pkg_install("Urban-Analytics-Technology-Platform/od2net/r")
 ```
 
 <div class="panel-tabset" group="language">
@@ -85,19 +107,24 @@ od_geo |>
 
 # od2net
 
-Building on code in the `od2net` and
-[nptscot/od2net-tests](https://github.com/nptscot/od2net-tests) repos,
-the code below prepares the input datasets and runs the
-network-generation code, generating output.geojson and output.pmtiles
-outputs:
+Building on code in the
+[od2net](https://urban-analytics-technology-platform.github.io/od2net/r/)
+R package, we can generate the files needed for the network generation
+stage.
+
+After you have installed the package, prepare the input datasets and run
+the network-generation code to generate output.geojson and
+output.pmtiles outputs:
 
 ``` r
-source("R/setup.R")
-make_zones("input/zones_york.geojson")
-make_osm()
-make_origins()
-make_elevation()
-destinations = destinations_york # Provided in the R package
+origin_zones = netgen::zones_york
+names(origin_zones)
+names(origin_zones)[1] = "name"
+sf::write_sf(zones, "input/zones_york.geojson", delete_dsn = TRUE)
+od2net::make_osm(zones_file = "input/zones_york.geojson")
+od2net::make_origins()
+netgen:::make_elevation()
+destinations = netgen::destinations_york # Provided in the R package
 names(destinations)[1] = "name"
 destinations = destinations[1]
 class(destinations$name) = "character"
@@ -125,9 +152,22 @@ fs::dir_tree("output")
 ```
 
     output
+    ├── counts.csv
     ├── output.geojson
+    ├── rnet.pmtiles
     └── rnet_output_osrm_overline.geojson
 
+From that point you can visualise the pmtiles in the website at
+[od2net.org](https://od2net.org) or other front-end application, as
+shown below.
+
+## Adjusting uptake functions
+
+The example above visualises *all* trips to school, not just active
+travel trips.
+
+## Other approaches to network generation
+
 An advantage of `od2net` is that it can generate the network and routes
 in a single step. For other approaches, we need to calculate the routes
 first, as shown below.
@@ -181,8 +221,6 @@ osmextract::oe_download(
 )
 ```
 
-    [1] "/home/robin/github/acteng/netgen/geofabrik_osm.pbf"
-
 Then with the system shell:
 
 ``` bash
@@ -262,5 +300,6 @@ plot(rnet)
 A disadvantage of this approach is that it’s computational
 resource-intensive and takes a long time. An in-progress is `od2net`.
 
-<!-- ## Comparing the outputs -->
 </details>
+
+# Comparing od2net and overline outputs

README.qmd

@@ -13,16 +13,32 @@ There are many sources of OD data, see [odgen](odgen.qmd) for a way to generate
 
 # Setup
 
+To run the code below you need to have R and Docker installed.
+After you have installed R, you can install the packages we'll use as follows:
+
 ```{r}
-#| include: false
-pkgs = c("sf", "tidyverse", "tmap")
+# Install pak if not already installed:
+if (!requireNamespace("pak", quietly = TRUE)) {
+  install.packages("pak")
+}
+pkgs = c("sf", "tidyverse", "tmap", "pak")
 pkgs_to_install = pkgs[!pkgs %in% installed.packages()]
 if (length(pkgs_to_install) > 0) {
-  install.packages(pkgs_to_install)
+  pak::pkg_install(pkgs_to_install)
 }
+# Load the packages with vapply:
+vapply(pkgs, require, logical(1), character.only = TRUE)
 ```
 
+We will also install a couple of package that are not on CRAN:
+
 ```{r}
+pak::pkg_install("acteng/netgen")
+pak::pkg_install("Urban-Analytics-Technology-Platform/od2net/r")
+```
+
+```{r}
+#| include: false
 # # Install geopandas:
 # reticulate::py_install("geopandas")
 ```
@@ -84,17 +100,21 @@ od_geo |>
 
 # od2net
 
-Building on code in the `od2net` and [nptscot/od2net-tests](https://github.com/nptscot/od2net-tests) repos, the code below prepares the input datasets and runs the network-generation code, generating output.geojson and output.pmtiles outputs:
+Building on code in the [od2net](https://urban-analytics-technology-platform.github.io/od2net/r/) R package, we can generate the files needed for the network generation stage.
+
+After you have installed the package, prepare the input datasets and run the network-generation code to generate output.geojson and output.pmtiles outputs:
 
 
 ```{r}
 #| eval: false
-source("R/setup.R")
-make_zones("input/zones_york.geojson")
-make_osm()
-make_origins()
-make_elevation()
-destinations = destinations_york # Provided in the R package
+origin_zones = netgen::zones_york
+names(origin_zones)
+names(origin_zones)[1] = "name"
+sf::write_sf(zones, "input/zones_york.geojson", delete_dsn = TRUE)
+od2net::make_osm(zones_file = "input/zones_york.geojson")
+od2net::make_origins()
+netgen:::make_elevation()
+destinations = netgen::destinations_york # Provided in the R package
 names(destinations)[1] = "name"
 destinations = destinations[1]
 class(destinations$name) = "character"
@@ -112,15 +132,7 @@ Run the tool with Docker as follows:
 ```{bash}
 #| eval: false
 # On Linux:
-sudo docker run -v $(pwd):/app ghcr.io/urban-analytics-technology-platform/od2net:main /app/config.json
-# or in Windows:
-sudo docker run -v ${pwd}:/app ghcr.io/urban-analytics-technology-platform/od2net:main /app/config.json
-```
-
-```{r}
-#| eval: false
-#| echo: false
-system("docker run -v $(pwd):/app ghcr.io/urban-analytics-technology-platform/od2net:main /app/config.json")
+sudo docker run -v $(pwd):/app ghcr.io/urban-analytics-technology-platform/od2net:main /app/config.json  --rng-seed 42 elevation_geotiff /app/input/elevation.tif
 ```
 
 After that you should see the following in the output folder:
@@ -129,6 +141,19 @@ After that you should see the following in the output folder:
 fs::dir_tree("output")
 ```
 
+From that point you can visualise the pmtiles in the website at [od2net.org](https://od2net.org) or other front-end application, as shown below.
+
+## Minimal od2net example
+
+For the purposes of testing it's worth 
+
+## Adjusting uptake functions
+
+The example above visualises *all* trips to school, not just active travel trips.
+
+
+## Other approaches to network generation
+
 An advantage of `od2net` is that it can generate the network and routes in a single step.
 For other approaches, we need to calculate the routes first, as shown below.
 
@@ -191,6 +216,7 @@ routes_osrm_minimal |>
 We can spin-up a local OSRM server to calculate routes as [follows](https://github.com/Project-OSRM/osrm-backend#using-docker):
 
 ```{r}
+#| eval: false
 location = osmextract::oe_match(
   od_geo_top_100 |> sf::st_union()
 )
@@ -285,6 +311,6 @@ plot(rnet)
 A disadvantage of this approach is that it's computational resource-intensive and takes a long time.
 An in-progress is `od2net`.
 
-<!-- ## Comparing the outputs -->
+</details>
 
-</details>
+# Comparing od2net and overline outputs

config.json

@@ -14,6 +14,5 @@
   },
   "cost": "Distance",
   "uptake": "Identity",
-  "lts": "BikeOttawa",
-  "elevation_geotiff": "elevation.tif"
+  "lts": "BikeOttawa"
 }