accessto is a library that allows calculation of access to opportunities, a commonly used measure in transportation planning and urban design. This library is designed to support analyses being conducted by City of Toronto staff, although others are welcome to use.
This package is designed so that users will use a Jupyter Noteook to setup and run the calculations.
Finding travel times and costs are the crucial underpinning to access to opportunities calcutions. This library calls two open source libraries that will actually perform these calculations OpenTripPlanner (OTP) and r5py. This package was developed using OTP v2.4, which was the newest stable release at the time of development.
This installation guide is designed primarily for City of Toronto employees to install this library and its dependencies on (Windows) City infrastructure. This may vary slightly for other users.
People working for the City of Toronto must use miniconda instead of the more complete-featured anaconda, as anaconda is not free for use by larger organizations.
The latest version of miniconda can be found here. https://docs.conda.io/projects/miniconda/en/latest/index.html
Install for a single-user (just me option), as to install for all users requires admin priviledges. At the time of writing, the default installation folder is C:\Users\[user name]\AppData\Local\miniconda3. You can use the default advanced installation options
From the Windows Start Menu, open *Anaconda Prompt (Miniconda3)
Create a conda environment that will hold rypy and its dependencies.
We had found an issue reading/writing shapefiles using the pyorgio package when we installed r5py directly using the instructions provided in the r5py installation guide. This affected pre and post-processing tasks, such as reading OSM networks. We found that installing the fiona and pyorgio packages before r5py appears to resolve this issue. We'd recommend also installing jupyterlab.
NOTE: No additional requirements are needed if using OpenTripPlanner to calculate travel times.
> conda create --name r5py python=3.11
> conda activate r5py
(r5py) [local directory]> conda install --channel conda-forge fiona pyogrio
(r5py) [local directory]> conda install --channel conda-forge r5py
(r5py) [local directory]> conda install --channel conda-forge jupyterlabNOTE: Note for City of Toronto users, that this will not work when using City desktop computers or when connected to the VPN. To run on a City computer I would install on a laptop not on the VPN, and then copy the Python environment to the desktop computer. Note that miniconda must be installed to the same location on both computers for this to work.
Download the code from this release and store in a convenient location. To download this code, click the green <> Code button at the top of the root page of this repository on GitHub, and either clone into a git repository, or download as a zip file, which you can extract.
NOTE: If using git on Windows on City of Toronto machines, we've found that we had to enter the following command in git before cloning the repository
git config --global http.sslbackend schannel. Thanks stackoverflow for the tip.
(r5py) [local directory]> pip install -e [path to the root folder of this library]Note that the root folder of the library contains the pyproject.toml file.
To update this package, replace the previous command with the following:
(r5py) [local directory]> pip install -e [path to the root folder of this library] --upgradeOpenTripPlanner is distributed as a single stand-alone runnable JAR file. OTP v2.5.0 can be downloaded from their GitHub repository. https://github.com/opentripplanner/OpenTripPlanner/releases
Download the 'shaded jar' file, e.g. otp-2.5.0-shaded.jar.
Both OpenTripPlanner and r5 use Java. If the system Java is too old, then you will to install a new JDK.
OpenTripPlanner version 2.5, requires JDK 17. The following JDK was used during development and testing of this library.
https://learn.microsoft.com/en-us/java/openjdk/download#openjdk-17
NOTE: JDK installation requires admin priviledges. City of Toronto staff will need to contact IT to install this package.
On City of Toronto computers, we've found that it's best to run using a local copy of the r5 .jar file.
Download the r5 JAR file, which is the package that performs the travel time computations. This is available on r5's GitHub page: https://github.com/conveyal/r5/releases.
As of the time of writing, the latest version is version 7.2, r5-v7.2-all.jar. Download this and store in a local directory.
You will then need to setup the r5py config file, which among other things will point r5py to this file. On a windows machine r5py looks for this file in the %APPDATA% \Roaming directory. You can find this by typing
%APPDATA% in the URL of an explorer window. Create a text file called r5py.yml in this directiory. The snippet below shows the contents of this file on the testing computer.
--max-memory=8G
--r5-classpath=C:\MyPrograms\r5\r5-v7.2-r5py-all.jar
--verbose=TrueMore information on r5py config files can be found on their documentation: https://r5py.readthedocs.io/en/stable/user-guide/user-manual/configuration.html#configuration-via-config-files
accessto is designed to be run from a Jupyter notebook. To open the notebook, perform the following steps:
- From the Windows Start Menu, open *Anaconda Prompt (Miniconda3)
- Activate the
r5pylibrary by typing the following in the Miniconda prompt window(r5py) [local directory]> conda activate r5py - Open a Jupyter Lab file, then open Jupyter lab
- first change directory to where your notebooks are stored
(r5py) [local directory]> cd [notebook location] (r5py) [notebook location]> jupyter lab
- first change directory to where your notebooks are stored
Several example scripts are included in the example_notebooks folder of this library package to help you get started.
Please also see the wiki for this project for more detailed documentation and usage guides.