Skip to content

byu-transpolab/wfrc_asim_scenario

Repository files navigation

WFRC ActivitySim Scenario

This repository includes activitysim as a submodule. This repository will need to be cloned with the --recurse-submodules flag to work correctly.

Instructions on how to run this scenario on the BYU supercomputer are given here.

This repository serves as a scenario builder for the ActivitySim implementation in the Wasatch Front / MAG modeling region. This is a research-only implementation to support work in travel modeling at BYU, and should not be used for any policy analysis.

This research is supported by the Utah Department of Transportation, and by US DOT via the T-SCORE University Transportation Center.

Environment Setup

The scenario builder is an implementation of the targets library for R. This is a make-style system that traces dependencies in a project and only re-builds if required.

Users should install the following R packages:

# only install if necessary

install.packages(c("targets", "tidyverse", "sf", "tigris", "tidycensus"))

Additionally, users should install Java 11, and python with the anaconda library. Users must also create an .Renviron file in the top-level folder with the path to java 11 as well as the API key to access the US Census bureau API.

JAVA_HOME="<path to java>/jdk-11.0.13.jdk/Contents/Home"

CENSUS_API_KEY="<your key here>"

Conda Setup

Activitysim and PopulatimSim both use conda environments to run their code. First conda will need to be downloaded and installed (miniconda3 works well), and then a few things must be done in a terminal. The scripts in this particular repository are run in bash, so it is best to run the following commands in a bash shell.

First verify that conda is in $PATH by running conda in a terminal. Then run conda init bash to initialize conda for a bash shell. This will add code to ~/.bashrc (or possibly ~/.bash_profile) that allows conda to work. Following that, restart the shell (or source ~/.bashrc and/or ~/.bash_profile) and create the required conda environment by running

conda env create --file=activitysim/conda-environments/activitysim-dev.yml \
--name ASIM_DEV

in a bash shell. This process may take quite a while (10-20 minutes). Conda should then be set up to work with this targets pipeline.

Input files

Some input files are included in this repository, if they come in under GitHub's 100 MB limit. Otherwise, targets will attempt to download the files from public folders on Box. The overall files are at this link.

Skims

The skims are derived from the 2019 forecast scenario (calibrated to 2015 data) run by WFRC / MAG with model version 8.3. The skims from this model run are stored in this box folder.

Because cube cannot run on Mac/Linux, we used the Cube Voyager script stored in sh/convert_cube_omx.s to convert these matrices to OMX and begin the targets pipeline with those OMX files, stored in a sister Box folder. The skims that are small enough to push to GitHub are included, the two skim sets that are too large to push are downloaded through targets.

  • skm_auto_Pk.mtx.omx
  • skm_auto_Ok.mtx.omx

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published