SiteSelection

SiteSelection aims to find complex zones to reallocate street space in the city, characterized by having limited and disputed spaces. These zones are also areas that have the potential for reallocating space dynamically over time, especially when multimodal demands are complementary.

It consists in a process of multi criteria to select the cell locations of a given city or neighbourhood where the street space is more disputed by different transport modes and street activities.

SiteSelection is a full script that uses a dynamic pipeline, and gathers and processes information on:

• Road network centrality measures (betweenness, closeness, degree)
• Population density
• POI and activities
• Public Transit Frequency
• Traffic levels (TBC)

Sources

The SiteSelection package is based in Portuguese open datasets, such as census and GTFS data.

Although it is easy to run for any location in Portugal, you may adapt the code to run at other locations (considering you have the similar data needed).

Data needed for other locations:

• Census data (population and buildings)
• POIs (see data_extract.R)
• GTFS data (see transit.R)
• Administrative boundaries

Setup

Installation

# install.packages("devtools")
devtools::install_github("u-shift/SiteSelection")

Requirements

• QGis and qgis_process installed and working
• targets R package.
• siteselection R package [under development].

Change defaults

Open the _targets.R file and change the defaults to your needs (don’t forget to save the file before run!):

# Set defaults HERE ######################
CITY_input = "Lisboa"       # Municipality name in Portugal
GEOJSON = FALSE             # use a different limit? made with https://geojson.io/ and saved in inputdata/*.geojson
GEOJSON_input = "map1"      # name of the file if GEOJSON = TRUE. default: "map1"
cellsize_input = c(600, 600)# in meters
square_input = TRUE         # TRUE = squares, FALSE = hexagons
use_h3 = FALSE              # use h3 to create universal grid?
h3_res = 8                  # h3 resolution. default: 9 (400m diameter). 8 = 1060m diameter, 10 = 150m diameter
build_osm = FALSE           # download and clean osm road network again?
analysis = TRUE            # export input parameters and results to a xls file? default: FALSE

# Thresholds for candidate selection
population_min = median     # mean or median? default: mean
degree_min = mean           # mean or median? default: mean
betweeness_range = 0.40     # percentile to exclude (upper and lower) default: 0.25
closeness_range = 0.25      # value to exclude (upper and lower) default: 0.25
entropy_min = 0.35          # value to exclude (lower) default: 0.5
freq_bus = c(4, 10, 20)     # frequency of bus stops to define classification. last 2 will remain. default: c(4, 10, 20) 

Run the pipeline

library(targets)
tar_visnetwork(targets_only = TRUE) # or false, to show objects

And you should have something like this

tar_make()
# let it run

tar_load(grid_all)
mapview::mapview(grid_all, zcol="score") #ranked cells

If you want to see the filtered cells (candidates), you can load the site_selection object and plot it, with complex and very complex cells (transit layer considered).

When the process is not complete, you may have an error like this

tar_visnetwork()

Work with other cities

You can change the city to any other in Portugal.

Other locations are work in Progress…

Work with non-administrative boundary

You can set another polygon limit to the analysis.
Just create a .geojson file with the area of interest and save it in the inputdata folder.
Then, set the GEOJSON = TRUE and GEOJSON_input to the name of the file, without the extension.

Use other grids

By default, the grid is created with squares. You can change it to hexagons by setting square_input = FALSE in _targets.R.

You can also use the universal h3 hexagon grid by setting use_h3 = TRUE and changing the resolution in h3_res.

Analysis

If you want to analyze the results in detail, set analysis = TRUE in _targets.R and load the analysis/analysis.table.Rds or .xlsx. It includes sequential runs with all inputs, outputs, and variable statistics.

About

Project

This work is part of Streets4All Project, developed at the University of Lisbon and at the University of Coimbra, and funded by Fundação para a Ciência e Tecnologia (PT).

The concept is based in:
Valença, G., Moura, F., & Morais de Sá, A. (2024). Where is it complex to reallocate road space? Environment and Planning B: Urban Analytics and City Science, 51(6). https://doi.org/10.1177/23998083231217770.

Presentation

This work was presented at the GET 2025 conference, January 2025.
Slides: https://u-shift.github.io/SiteSelection/slides

Cite

Please refer to this work as:
Félix R, Valença G (2024). SiteSelection: An R script to find complex areas for the Streets4All Project. R package version 0.2, https://github.com/U-Shift/SiteSelection/.