README.md
In tripaccess: American Travel Behavior and Access Datasets

tripaccess

This data package contains four datasets tripaccess, person, house, and trip, constructed from the National Household Travel Survey (NHTS) 2017 person, house, and trip files (Federal Highway Administration, 2017). It includes personal trips, mobility, demographic, and household information. It is suitable for data visualization, data wrangling, joining datasets, exploratory data analysis (EDA), group comparisons, simple linear regression, categorical data analysis, and data ethics discussion in data science and statistics classes.

One goal of this data package is to increase awareness of disability inclusion by examining factors or characteristics that are associated with different travel behaviors of people who have a travel disability. The tripaccess and person datasets include a variable travel_disability, indicating whether respondents have a medical condition “that makes it difficult to travel outside of home”. These two datasets also include travel accommodations information, e.g., walker, cane.

Note that the NHTS 2017 has imbalanced classes of people who have a travel disability and people who do not have a travel disability. So this data package can be used for data ethics discussion, e.g., ethical concerns of underrepresentation of marginalized groups in your data.

The transport system is a pillar for ensuring social equity (Pagliara & Di Ciommo, 2020). People often need to travel to work, study, connect with other people, shop groceries, attend medical appointments, and participate in fun events. There have been uneven distributions of travel resources in the current built environment for disabled people, which causes barriers to access for them (Levine & Karner, 2023). This data package can be used to study economic and social participation from a critical disability lens and provide policy implications for building a more equitable and inclusive transport system.

Install tripaccess from CRAN:

install.packages("tripaccess")

Install the development version of tripaccess from GitHub:

# If you haven't installed the pak package yet, do so:
# install.packages("pak")
pak::pak("scao53/tripaccess")

# Load package
library(tripaccess)

tripaccess: tripaccess, constructed from the NHTS 2017 person and trip files, includes disability and other demographic as well as mobility and trip categorical and numeric variables. It has 86,521 rows (each row is a person, filtered to age 18-61) and 41 columns. It includes person weights that can be used for person-level estimates, e.g., workers by gender, drivers by annual miles estimated, etc. It can be used for advanced exercises, e.g., data visualization, single table data wrangling, EDA, group comparisons, and categorical data analysis.
person: person, constructed from the NHTS 2017 person file, includes disability, mobility, and other demographic categorical and numeric variables. It has 99,564 rows (each row is a person, filtered to age 18-61) and 32 columns. It can be used for advanced exercises, e.g., data visualization, single table analysis, EDA, group comparisons, and categorical data analysis. It can also be used together with the house and trip datasets for joining datasets (see Data Relationships).
house: house, constructed from the NHTS 2017 house file, includes household characteristics categorical and numeric variables. It has 129,695 rows (each row is a household) and 9 columns. It can be used for beginners’ exercises, e.g., data visualization, data wrangling, and EDA. It can also be used together with the person and trip datasets for joining datasets (see Data Relationships).
trip: trip, constructed from the NHTS 2017 trip file, includes trip related categorical and numeric variables. It has 921,590 rows (each row is a trip) and 8 columns. It can be used for beginners’ exercises, e.g., data visualization, data wrangling, EDA, and simple linear regression. It can also be used together with the person and house datasets for joining datasets (see Data Relationships).

The person, house, and trip datasets can be used together for joining data. The diagram below shows how these three datasets can be joined by different key variables. The top of each table shows the dataset name. The grey-shaded variables are primary keys while the others are foreign keys. The arrows show how the datasets are connected.

Use person_id and household_id to join the person dataset and the trip dataset.
Use household_id to join the house dataset and the person dataset.
Use household_id to join the house dataset and the trip dataset.

Here is an example of EDA and group comparisons using the person dataset.

library(tripaccess)
library(tidyverse)

#> Summary statistics of public transit use by travel disability status
transit_summary <- person |>
   group_by(travel_disability) |>
   summarize(
     people = n(),
     public_transit_users = sum(count_of_public_transit_usage > 0),
     public_transit_use_prop = mean(count_of_public_transit_usage > 0),
   )
transit_summary
#> # A tibble: 2 × 7
#>   travel_disability_sort_val people public_transit_users public_transit_use_prop
#>   <fct>                       <int>                <int>                   <dbl>
#> 1 No Travel Disability        92897                12190                   0.131
#> 2 Travel Disability            6667                 1213                   0.182

#> Test whether public transit use differs by travel disability status
prop.test(
   x = transit_summary$public_transit_users,
   n = transit_summary$people
#>  2-sample test for equality of proportions with continuity correction
#> 
#> data:  transit_summary$public_transit_users out of transit_summary$people
#> X-squared = 136.93, df = 1, p-value < 2.2e-16
#> alternative hypothesis: two.sided
#> 95 percent confidence interval:
#>  -0.06031243 -0.04112818
#> sample estimates:
#>    prop 1    prop 2 
#> 0.1312206 0.1819409

Here is an example of data wrangling and visualization using the house dataset.

library(tripaccess)
library(tidyverse)

#> Filtered to households with at least one driver
house_with_drivers <- house |>
   filter(number_drivers > 0)

#> Filtered to households with at least one vehicle
house_with_vehicles <- house_with_drivers |>
   filter(number_vehicles > 0)

#> Plot household vehicles by number of drivers
ggplot(data = house_with_vehicles,
       aes(x = number_drivers,
           y = number_vehicles)) +
   geom_jitter(alpha = 0.08, width = 0.15, height = 0.15) +
   geom_smooth(method = lm, se = FALSE, formula = y ~ x, color = "blue") +
   labs(title = "Household Vehicles versus Number of Drivers",
        x = "Number of Drivers in Household",
        y = "Number of Household Vehicles") +
   theme_bw()

For more example usage of the tripaccess, person, house, and trip datasets, please view the vignette.

The person, house, and trip datasets are subsets of the NHTS data and unweighted. The results of your analyses cannot be generalized beyond the samples collected in these datasets. The tripaccess dataset, as a subset of the NHTS data, includes person weights that can be used for person-level estimates.

Federal Highway Administration. (2017). 2017 National Household Travel Survey, U.S. Department of Transportation, Washington, DC. Available online: https://nhts.ornl.gov.

Levine, K., & Karner, A. (2023). Approaching accessibility: Four opportunities to address the needs of disabled people in transportation planning in the United States. Transport Policy, 131, 66-74.

Pagliara, F., & Di Ciommo, F. (2020). Editorial. Transportation Research Part A, 141, 396-397.