Skip to content

Data and Methodologies: Frequently Asked Questions

Data Description

1. How are the data collected? Where do these data come from?

LOCUS Truck data originates from truck telematics data derived from in-vehicle systems that track all movements of a given truck. We work with the largest vendor of in-vehicle truck telematics receivers in the US. Data are collected whenever the engine of the vehicle is turned on with location information updated every one second, which ensures a robust stream of locational information capable of tracking a vehicle at specific stop locations as well as over the highway network. However, in order to protect privacy and contractual considerations of the companies and fleets it works with, our vendor does not provide this level of granularity in spatial and temporal travel patterns. Q-4 addresses the level of geographic scale available in LOCUS Truck.

2. What are the vehicle/truck types in the dataset?

There are three truck types included in the dataset. - Light Duty Truck (LDT) – All commercial trucks of Gross Vehicle Weight Rating (GVWR) 10,000 lb or below (Federal Highway Administration “FHA” class 1-2) are included in this category. This category includes both light duty trucks (LDT) and multipurpose vehicles (MPV). - Medium Duty Truck (MDT) – All commercial vehicles of GVWR between 10,001 lb and 26,000 lb (FHA class 3-6) are included in this category - Heavy Duty Truck (HDT) – All commercial vehicles of GVWR between 26,001 lb or above (FHA class 7-8) are included in this category.

3. What is the minimum stop duration considered for a trip to end?

A truck trip is considered to have ended if either of the following conditions are met: - The truck engine is turned off. - The truck engine is not turned off, but the truck is idle (no movement) for a specified duration. - The duration threshold used for the second condition depends on the truck size. For Light-Duty Truck, it is 5 minutes; for Medium-Duty Trucks, it is 15 minutes; and for Heavy-Duty Truck, it is 30 minutes.

4. What is the standard geographic scale/resolution?

While our vendor allows the telematics data to be queried at any level of aggregation, privacy filters used in the process result in data loss when granularity is high. We have found that the Census Tract level of geography provides the best middle ground that ensures minimal data loss while providing meaningful travel pattern trends of trucks by size. In addition to geographic segmentation, LOCUS Truck product provides segmentation along several other dimensions including (1) time of day, (2) truck size, (3) industry class, (4) vocation. The GPS data that are used in the expansion process are aggregated at the tract level. The standard product includes tract-level detail for truck OD flows. The LOCUS Truck UI also allows for custom, user-defined zone systems as long as the zone systems are no more granular than Census Tracts.

5. What is the sample size of the truck GPS data?

We find that the sample penetration rates, when measured against traffic counts, are typically in the range of 8 to 10 percent, though results vary by region, truck type, and year of the data.

6. What are the known biases in the data samples?

Sample loss due to data privacy and biases to short-distance trips

To safeguard the privacy of individual driver, trip, and fleet information, our data vendor provides us with zonal-level (i.e., tract level) aggregated truck GPS sample data. Our data vendor applies a minimum threshold for aggregating trips at the zonal level OD pairs. If the sample size is below the threshold, samples for those zonal OD pairs are filtered out from the sample dataset provided by our vendor. This causes filtering out of long-distance OD pair trips more than the short-distance ones, causing the sample data to be biased toward short-distance trips. We measure the sample loss at different geography levels and trip lengths and incorporate methods in our expansion process to eliminate the impact of these data filtering biases in our expanded data product.

Not capturing short-duration stops in sample data

Due to the minimum stop duration thresholds discussed above, some legitimate short-duration stops are not captured by our sample data when the truck’s engine is not turned off. This is a tradeoff made to avoid added noise that would result from the appearance of short-duration stops caused by traffic control devices or congestion, which are not legitimate stops for business.

An example of such legitimate stops that might not be captured are stops made by door-to-door package delivery vehicles that may make stops within a neighborhood to deliver a package without turning the engine off. To reduce the biases introduced by the minimum stop duration thresholds, we use trajectory data from additional data sources to estimate the number of door-to-door delivery trips made into, out of, and within a given Tract.

7. How are the data expanded to the population?

We use a behaviorally-based approach to expand truck data to the population that relies on a collection of commercial vehicle surveys to develop expansion targets at the Census Tract level across the US. Expansion targets rely primarily on the employment characteristics of the tract, but also consider key special generators like ports, intermodal facilities, truck parking locations, and warehouse locations. Expanded trip totals found in the product represent estimates of an average day over a particular calendar year.

8. Does the dataset capture e-commerce related delivery trips?

Our Light Duty Truck trip dataset includes both ecommerce related door-to-door delivery trips and other service-related trips. In addition to the trip expansion approach discussed above in Q-7, for ecommerce related door-to-door delivery trip expansion, we used additional package delivery survey and package delivery route trajectory data to develop the expansion targets and also capture the travel pattern of such trips.

9. How are trip patterns validated?

We validate our expanded truck travel pattern data against established, standardized external data sources such as the National Vehicle Inventory and Use Survey (VIUS), which provides estimates of the truck inventory as well as truck VMT estimates; commercial vehicle surveys which can provide key distributions like trip lengths and time of day characteristics; traffic count data along major corridors; and truck model systems (which are typically validated to traffic counts locally). In addition, we use a variety of sense checks to ensure that the patterns in the data are meaningful and reasonable.