Broadening Understanding of the Interplay Among Public Transit, Shared Mobility, and Personal Automobiles (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

35 A major transportation network company (TNC) provided the researchers with data about trip origin and destination, aggregated by zip code tabulation area (ZCTA) and hour of day, for the month of May 2016 in five different U.S. regions. While just a small sample of overall TNC activity nationwide, the data presents a window into TNC behaviors in a range of regional contexts. This section contains a description of the TNC dataset, and the following section comprises statistical overviews of the activity in each region. Data covers the following regions, for trips originating in the counties containing or adjacent to each region’s core city (listed in parentheses): • Chicago (Cook County, Illinois), • Los Angeles (Los Angeles and Orange counties, California), • Nashville (Davidson County, Tennessee), • Seattle (King County, Washington), • Washington, D.C. (District of Columbia; Montgomery and Prince George’s counties, Maryland; Alexandria, Arlington, and Fairfax counties, Virginia) Data Format A major TNC provided the researchers with data about all trips originating in five metro­ politan areas during the month of May 2016. There was one dataset for each metropolitan area, with each dataset formatted in traditional origin­destination design and containing five variables: origin zip code, destination zip code, day of week of trip start, hour of trip start, and a measure called “percentage.” Each row of the data represents a unique origin­destination zip code pair for a single hour in the week, with a single value for that sample. A given zip code pair can be represented in up to 168 rows (the product of seven 24­hour days). The same zip can represent both origin and destination, which is the case for more than one in eight significant flows. Origin­destination pairs with zero rides in a given hour of the week are not represented in the data. The percentage column represents the number of rides in May 2016 aggregated into a single week, indexed to the highest ride volume of destination­origin­time combination per market. Values range from 0 to 100, with 100 representing the trip volume of the single largest flow (i.e., the origin­destination pair with the most trips during a single hour) encountered in that market in the sample week. Each market has its own 0 to 100 range based on the volumes in that market—so these figures cannot be directly compared across markets. That is, a flow with a percentage 90 in Los Angeles does not necessarily represent the same actual volume as a flow with percentage 90 in Washington, D.C. A P P E N D I X A Transportation Network Company Trip Data Overview

36 Broadening Understanding of the Interplay Among Public Transit, Shared Mobility, and Personal Automobiles Percentage values between 0 and 2 are represented in the data by the text “insignificantly small” in the original dataset, which indicates that there was some flow during that hour, but not one for which we have any representation of a volume. This feature of the data permits a distinction between flows that are nonzero but insignificant, and flows that are nonexistent. Often a given ZCTA pair will appear in the data only sporadically, for the few times over the course of the week that it had flows. Since the relative magnitude of the insignificant flows is impossible to know beyond the fact that they are nonzero, they are included only in counts where volume is not a factor. For calcula­ tions that rely on volume (such as for trip length, which uses volume­weighted distances), the insignificant flows are dropped. For each region, the dataset only includes trips originating in zip codes in the county containing the core city of the region, but will show any destination zips regardless of location—in a few cases the destinations are several states away from the origins, though these are most likely one­off occurrences. In the Los Angeles region, the data includes origins in both Los Angeles and Orange counties. In the Capital region, the data includes origins in Washington, D.C., as well as in adjacent Virginia and Maryland. Statistical Description The trip length estimation (Table A­2) is based on the straight­line distance between the geographical center points (centroids) of the origin and destination ZCTAs. For trips within a single ZCTA, the distance used is one­half of the longest possible straight­line trip within the area, which is essentially the average of the shortest and longest possible trips within that ZCTA. A trip over city streets between the same two points would in almost every case be longer, though trips between two ZCTAs might be much longer or shorter than this distance depending Chicago, IL Los Angeles, CA Nashville, TN Seattle, WA Washington, D.C. Observations 231,944 753,736 33,972 76,166 241,592 Core county ZCTAs 188 416 37 89 225 Unique origin ZCTAs 169 359 32 73 156 Unique destination ZCTAs 311 547 61 135 336 Unique O/D pairs 13,937 50,510 1,052 3,631 13,623 Table A-1. Statistical overview of the study regions. Region Minimum 25th percentile Median Mean 75th percentile Maximum Average transit trip1 Chicago 0.35 1.42 2.32 2.88 3.37 21.36 4.35 D.C. 0.18 1.61 2.24 2.60 3.11 28.02 4.79 LA 0.36 1.58 2.42 2.97 3.60 30.67 4.88 Nashville 0.33 1.80 3.06 3.87 5.29 20.36 5.39 Seattle 0.69 1.69 2.36 3.02 3.47 21.52 4.88 Table A-2. Range of regional TNC trip lengths, and average regional transit trip length (miles). 1Average transit trip lengths are calculated from the ratio of passenger miles to unlinked passenger trips on the major transit agencies of the region, excluding commuter rail. Source: 2016 APTA Factbook, Appendix B.

Transportation Network Company Trip Data Overview 37 on exactly where within each ZCTA the origin and destination points lie. Since we do not have access to this level of geographic information, the straight­line distance between centroids pro­ vides a readily calculated and reproducible proxy that gives a sense of the relative magnitude of distances in familiar units. But because this measure underestimates the actual driving distance between the two points, the figures produced should be regarded as lower­bound estimates of the actual distance. The aggregate statistics shown in these tables are weighted by trip volume, and do not account for flows for which the data did not supply a trip volume (i.e., flows below 2% of the highest volume).