A Large-scale Dataset for Commuting OD Matrix Generation

This dataset consists of a total count of 3,233 areas around the United States, using counties and metropolitons as area boundaries as well as census tracts and CBGs as region units in the corresponding area, each area including its regional spatial characteristics and commuting OD matrix.

1. Data Description

Regional Spatial Characteristics

Each region is characterized by demographics and urban functionalities, derived from American Community Survey (ACS) by the U.S. Census Bureau and the distribution of POIs from OpenStreetMap. Demographics include the population structure of a region based on age, gender, income, education, and other factors, encompassing a total of 97 dimensions. POIs are divided into 36 different categories. The distances between regions are calculated using the planar Euclidean distance between their centroids.

Characteristics of Demographic Part
Total Population
Male Population
Female Population
Under 5 Years - Total
Under 5 Years - Male
Under 5 Years - Female
5 to 9 Years - Total
5 to 9 Years - Male
5 to 9 Years - Female
10 to 14 Years - Total
10 to 14 Years - Male
10 to 14 Years - Female
15 to 19 Years - Total
15 to 19 Years - Male
15 to 19 Years - Female
20 to 24 Years - Total
20 to 24 Years - Male
20 to 24 Years - Female
25 to 29 Years - Total
25 to 29 Years - Male
25 to 29 Years - Female
30 to 34 Years - Total
30 to 34 Years - Male
30 to 34 Years - Female
35 to 39 Years - Total
35 to 39 Years - Male
35 to 39 Years - Female
40 to 44 Years - Total
40 to 44 Years - Male
40 to 44 Years - Female
45 to 49 Years - Total
45 to 49 Years - Male
45 to 49 Years - Female
50 to 54 Years - Total
50 to 54 Years - Male
50 to 54 Years - Female
55 to 59 Years - Total
55 to 59 Years - Male
55 to 59 Years - Female
60 to 64 Years - Total
60 to 64 Years - Male
60 to 64 Years - Female
65 to 69 Years - Total
65 to 69 Years - Male
65 to 69 Years - Female
70 to 74 Years - Total
70 to 74 Years - Male
70 to 74 Years - Female
75 to 79 Years - Total
75 to 79 Years - Male
75 to 79 Years - Female
80 to 84 Years - Total
80 to 84 Years - Male
80 to 84 Years - Female
85 Years and Over - Total
85 Years and Over - Male
85 Years and Over - Female
Median Age - Total
Median Age - Male
Median Age - Female
Median Earnings (Dollars)
Class of Worker - Private Wage and Salary Workers
Class of Worker - Government Workers
Class of Worker - Self-Employed Workers
Class of Worker - Unpaid Family Workers
Travel Time to Work - Mean Travel Time (Minutes)
Vehicles Available - No Vehicle Available
Vehicles Available - 1 Vehicle Available
Vehicles Available - 2 Vehicles Available
Vehicles Available - 3 or More Vehicles Available
Total Households
Average Household Size
Total Families
Average Family Size
Nursery School, Preschool
Kindergarten to 12th Grade
Kindergarten
Elementary: Grade 1 to Grade 4
Elementary: Grade 5 to Grade 8
High School: Grade 9 to Grade 12
College, Undergraduate
Graduate, Professional School
9th to 12th Grade, No Diploma
Associate's Degree
Bachelor's Degree
Bachelor's Degree or Higher
Graduate or Professional Degree
High School Graduate (Includes Equivalency)
High School Graduate or Higher
Less Than 9th Grade
Less Than High School Graduate
Population 25 to 34 Years - Bachelor's Degree or Higher
Population 25 to 34 Years - High School Graduate or Higher
Some College or Associate's Degree
Some College, No Degree
Poverty - Male
Poverty - Female

Characteristics of Point-of-interests Part
finance
public
transport
entertainment
health
service
education
government
religion
accommodation
food
cafe
fast_food
ice_cream
pub
restaurant
shop_beauty
shop_clothes
boutique
shop_transport
retail
commodity
marketplace
home-improvement
sport
public_transport
kindergarten
office
recycling
travel_agency
tourism
shop_livelihood
residential
dormitory

Commuting OD Matrices

We construct the OD matrices for all areas using data on commuting patterns from the 2018 Longitudinal Employer-Household Dynamics Origin-Destination Employment Statistics (LODES) dataset. These matrices represent aggregated commuting flows within areas. Each entry in an OD matrix denotes the count of individuals residing in one region and working in another, effectively mapping the commuting patterns of workers across different regions. The LODES dataset is widely used in existing works. In this dataset, the commuting information is aggregated by the cooperation and other kind of work units, which is more reliable and accurate than the individual commuting data. Therefore, in the data collection process, information has been ensured to be representative at a national scale, thus eliminating sampling errors.

2. Benchmark Experiments

Prerequisites

To conduct the benchmark experiments, users should first prepare a Python environment containing the necessary libraries.

To conduct the experiments, ensure you have the following setup:

• Python Version: Python 3.8

• Required Libraries:

  • numpy
  • scikit-learn==1.3.0
  • torch==2.1.0+cu118
  • scipy==1.10.1
  • dgl==1.1.2+cu117
  • networkx==3.1

Please install or update these libraries to match the specified versions for optimal compatibility.

Run Experiments

The data of metropolitans can be found at this link.

The code for the benchmark experiments is stored in the ./model/* directory. Each model-specific folder contains a main.py file, which serves as the entry point for executing the benchmark experiment. To run a specific benchmark experiment, simply execute the main.py file directly.

To run the model, follow these specific steps:

1. Navigate to the Project Root Directory:

   Use the cd command to change into the project's root directory.

   cd path_to_this_proj

2. Execute the Model Script:

   Run the main.py file located in the WeDAN model directory ./WeDAN by using the following command:

   python WeDAN/main.py

The experimental results are shown as follows.

Model	CPC	RMSE	NRMSE	inflow	outflow	ODflow
GM-P	0.321	174.0	2.222	0.668	0.656	0.409
GM-E	0.329	162.9	2.080	0.652	0.637	0.422
SVR	0.420	95.4	1.218	0.417	0.555	0.410
RF	0.458	100.4	1.282	0.424	0.503	0.219
GBRT	0.461	91.0	1.620	0.424	0.491	0.233
DGM	0.431	92.9	1.186	0.469	0.561	0.230
GMEL	0.440	94.3	1.204	0.445	0.355	0.207
NetGAN	0.487	89.1	1.138	0.429	0.354	0.191
DiffODGen	0.532	74.6	0.953	0.324	0.270	0.149
WeDAN	0.593	68.6	0.876	0.291	0.269	0.147

Extra Usage of This Dataset

To validate your model using this dataset, you can utilize existing scripts for data loading and performance evaluation. Follow these steps to set up and execute your experiments:

1. Create a Directory for Your Model:

   Prepare a folder to house your model experiments:

   mkdir ./name_of_your_model

2. Write Your Model and Entry Script:

   • Develop your own model.py to describe your model.
   • Create a main.py to serve as the entry point of your experiment. This script should reuse the existing data_load.py and metrics.py:

3. Execute Your Model:

   Navigate to the project root directory and run your model:

   cd path_to_this_proj
   python ./name_of_your_model/main.py