1. Executive summary

This document reports the accuracy of the TomTom Historical Traffic Volumes hourly estimates for 2026, the current year. Sections 2 to 4 describe the product and the metrics and are the same in every year’s document; Section 5 describes how the current year is validated, and Section 6 and the key findings below are specific to 2026.

We measured accuracy against independent ground-truth traffic counts from permanent loop detectors and similar counting infrastructure in five countries (Belgium, Netherlands, Norway, United Kingdom, and United States) for 2026. The per-country metrics for 2026 come from a direct comparison of the product with the loop counts recorded in 2026, the freshest ground truth available (Section 5). Past years are validated by k-fold cross-validation over the full set of counters; Section 5 explains the difference. A market is a country where the product is available.

Key findings:

• The United Kingdom delivers the most accurate hourly estimates in this comparison: a MAPE of 4.6% on high-volume roads and 7.7% on medium-volume roads in day hours, with a median percentage error of 0.1% and -0.1% respectively, so the estimates carry no measurable bias.
• Across the five countries, the day-hour MAPE on medium-volume roads ranges from 7.7% (United Kingdom) to 17.4% (Norway); in Belgium, the Netherlands, the United Kingdom and the United States every road category in both day and night hours reaches at least the Acceptable SQV tier, and most reach Good or Very good.
• Norway shows a consistent tendency to estimate above the counts: the median percentage error is positive in every road category, from 9.2% on medium-volume roads in day hours to 28.6% on low-volume roads at night, and it has no counters on high-volume roads.
• As in every year, percentage errors are highest on low-volume roads and in night hours, when a difference of a few vehicles per hour is a large share of the count: the MAE on low-volume roads at night is under 11 vehicles per hour in every country. Section 7 explains how to read these figures for a given use case.

2. Introduction

How many vehicles pass a road segment in a given hour? TomTom Historical Traffic Volumes estimates exactly that, for each road segment. We apply machine learning to probe data: the position and speed reports that millions of connected vehicles send (Sekuła et al., 2018; Zhan et al., 2017). The estimates for individual hours cover the roads with significant probe coverage, by road class and country (Section 3.5). A road class is the functional class of a road in the map, such as motorway, major road or local street.

Historical Traffic Volumes delivers three quantities: AADT, AAWHT and hourly volumes. The product introduction defines them. This document covers the hourly volumes; AADT and AAWHT are documented in the Annual averages section of this documentation.

The hour-by-hour estimates reach back to 2024 and forward to about 72 hours before the present. We improve the model continuously, and an improved model can recompute past periods, so a published figure for a past period can improve after publication (Section 8).

Traffic volume data informs decisions with real consequences:

• Where to open a new store
• How to allocate infrastructure budgets
• How to assess road safety risk
• Whether the surrounding road network can support a proposed development
• Whether a new lane, a closure or another change to a road altered traffic, comparing the period before with the period after

Every one of those decisions rests on a volume number, so the useful question is not whether the data is reliable but how far off it can be on the roads you care about. This document answers that in vehicles, in percentages and as quality scores.

Traditional counting methods give precise counts at a small number of fixed points (Federal Highway Administration, 2022). TomTom Historical Traffic Volumes estimates the volume across the road network of a covered market (Section 3.5 gives the coverage of the hourly estimates). An estimate carries uncertainty, so we measure that uncertainty with the metrics defined in Section 4. With those numbers, customers can decide where and how to use the data.

We validate against counters: permanent loop detectors and similar counting stations, independent of the model. The accuracy metrics in this document measure the gap between our estimates and those reference counts; a smaller gap means a more accurate estimate. Where results fall short of the quality thresholds in Section 4.4, we say so and give the available context.

3. How the model works

Turning raw reports from connected vehicles into reliable, network-wide volume estimates takes a sequence of steps, and each step solves a distinct problem. Unlike traditional counting methods, the product needs no physical equipment at each measurement point. We collect probe observations (Section 3.1), estimate the penetration rate (Section 3.2) and convert probe observations into volume estimates: first the typical values (Section 3.3), then the values for specific hours (Section 3.4). Section 3.5 states where the hourly estimates are available.

3.1 From connected vehicles to traffic observations

Our primary data source is floating-car data: GPS and telematics signals from connected vehicles (Herrera et al., 2010). Each signal is a probe observation; together, the signals form the probe data. Connected vehicles include:

• In-vehicle navigation systems
• Smartphones running navigation apps
• Connected commercial vehicles

The signals arrive passively and continuously from millions of devices worldwide. For each road segment, they form a continuous stream of speed and passage observations.

Probe observations are not traffic volumes, though. Only a fraction of the vehicles on a given road are connected and contributing data. We call this fraction the penetration rate. It varies by road type, geography and time of day, so converting probe observations into total volume estimates means accounting for that variation.

3.2 Estimating the penetration rate without counters

A probe count becomes a traffic volume only once we know the penetration rate: the share of vehicles on that road that report probe data. Permanent counters measure it directly, but they exist on a small fraction of roads, and in many markets on none. A method that needs counters everywhere cannot scale. Ours needs them only once, to learn.

The key is congestion. When a road operates at or near capacity (Transportation Research Board, 2022), physics constrains it: the relationship between the speed vehicles drive and the number of vehicles the road carries becomes tight and predictable, as the fundamental diagram of traffic flow describes (Greenshields, 1935; Treiber & Kesting, 2013). From probe speeds, read together with the road’s attributes, we can then estimate how many vehicles the road carries. We train a capacity model on roads with counters to learn this speed-to-flow relationship, together with map attributes such as road class, urban or rural context and lane configuration. Because the same congestion physics applies wherever a road runs near capacity (Kerner, 2004), the model transfers to roads that have never had a counter.

On any congested road we then hold two independent numbers: the total flow the capacity model estimates from speeds, and the probe flow we count directly. Their ratio is the penetration rate. Wherever probes meet congestion, we obtain a penetration-rate estimate, far beyond counter locations (Eisinga & Lorkowski, 2025). Individual estimates are noisy, so we aggregate them by region and road type into summaries that resist noise and fill the remaining gaps from similar surroundings. The result is a consistent picture of probe representativeness across countries and road classes. Roads that never congest inherit the estimate of their region and road type.

3.3 From penetration rate to typical volumes

The regional penetration picture tells us roughly what share of traffic the probes capture around a road. The volume model turns that into an estimate for the specific road. We train it on real ground truth: permanent counters, where they exist. From those counts it learns how the regional penetration rate, the observed probe data and the road’s own attributes combine into the volume of one road, in effect refining the regional penetration rate down to each segment. Once trained, it needs no counters. It runs wherever probe data and a map exist, which is what makes the product scalable to new regions.

A motorway and a residential street sit at opposite ends of a wide range: among the counters the model learns from, the quietest carry fewer than 500 vehicles a day and the busiest more than 125,000. The relationship between a road’s attributes, its probe activity and its traffic load is not the same at the two ends. The estimate has to hold across that whole range. So we train the model on counters across the full range of volumes, and we judge it on the same range, separately for high-, medium- and low-volume roads in every validated country. The Annual averages section of this documentation reports those results for AADT; Section 6 of this document reports them for the hourly estimates.

The volume model produces the typical values first: AADT, annual average daily traffic, one value per road segment and year, and AAWHT, annual average week-hour traffic, 168 values per segment and year, one for each hour of each day of the week. Section 3.4 describes how the estimate for one specific hour builds on them.

3.4 From typical hours to specific hours

AADT and AAWHT describe typical conditions. Customers usually want to know about one road on one day at one hour — last Tuesday at 8am, for example.

The model produces these estimates by starting from the typical value for that road, that day of the week and that hour, and adjusting it by how busy the road was in the period being estimated, as observed in probe data. A road that was quieter than on a typical Tuesday morning receives a lower estimate; a road that was busier receives a higher one.

The set of vehicles contributing probe data is not fixed: sources are added and removed over time, and the number of reporting vehicles varies from road to road. A method that read absolute probe counts would mistake fewer reporting vehicles for less traffic. The model therefore measures the adjustment as a relative change in probe activity: it compares the period being estimated with a reference measured on the same basis as that period, so that a change in the number of reporting vehicles does not read as a change in traffic.

Where too few vehicles report on a road class in a country for hourly estimates to be reliable, the model produces no hourly estimate; Section 3.5 describes the coverage of the hourly estimates.

3.5 Coverage of the hourly estimates

TomTom Historical Traffic Volumes has two products with two footprints. The annual averages, AADT and AAWHT, cover most roads of the network. The hourly estimates described in Section 3.4 need enough reporting vehicles, so they cover the roads with significant probe coverage, by road class and country. Where too few vehicles report on a road class in a country for the hourly estimates to be reliable, we publish none for it. This is a deliberate choice: a published but unreliable estimate would be used as if it were sound, and we prefer a gap to a number that looks sound but is not.

Coverage therefore differs by road class within a country. Larger roads carry more traffic, and with it more reporting vehicles, so hourly estimates are available on major roads in more countries than on minor roads. The Market coverage page of the Hourly section lists, for each country, the road classes with hourly estimates.

That list grows. We are extending the hourly model to the roads it does not yet cover: roads where low volumes mean few reporting vehicles and a sampling error in the probe signal. We improve the model continuously, and further releases also regenerate historical data, so coverage and accuracy improve for past periods too. The Market coverage page shows the current state. Earlier periods can carry hourly estimates on roads that the current list does not include.

4. How we measure accuracy

One number cannot show both the typical error and its spread, so we report a set of complementary metrics. Each one highlights a different aspect of performance, and every one is measured against independent ground-truth counts. The unit of analysis is the segment-hour: one road segment in one hour on one date. Every metric in this document is computed across segment-hours, so a road segment contributes one value per counted hour, not one value per segment.

4.1 Mean absolute percentage error (MAPE)

MAPE measures the average size of the prediction error relative to the observed count, as a percentage. A MAPE of 10% means that estimates differ from actual counts by 10% on average. MAPE treats errors of all sizes equally, and it is a commonly reported accuracy measure in traffic estimation.

On quiet roads and in quiet hours, a high percentage error can mean a small difference in vehicles (Hyndman & Koehler, 2006). A road carrying 40 vehicles in an hour, with a MAPE of 15%, has a typical absolute error of 6 vehicles in that hour. For most planning and analytical purposes, a difference of that size is negligible. This is one reason Section 6 reports day and night hours separately: night volumes are low, so percentage errors rise, while the MAE column shows the error in vehicles. So read MAPE values for low-volume roads alongside the absolute volumes that matter for your use case.

4.2 Median percentage error

The median percentage error is the middle value of all signed errors: positive when our estimate exceeds the actual count, negative when it falls short. A value close to zero shows that the model has no strong tendency to over- or under-count. Low bias matters for applications such as aggregated network analysis or vehicle kilometers traveled (VKT) calculations, because systematic errors add up across many road segments.

4.3 68th and 95th percentile absolute percentage error

These percentiles describe how the errors spread across segment-hours. The 68th percentile corresponds roughly to one standard deviation in a normal distribution: about 68% of segment-hours have an error at or below it. The 95th percentile captures the upper range, the error level of the most challenging segment-hours. Together with MAPE, the percentiles show how the error is distributed, not only its average.

4.4 Scalable Quality Value (SQV)

A percentage error looks large on a quiet road and an absolute error looks large on a busy one. The Scalable Quality Value (Friedrich et al., 2019) handles both cases. It generalizes the GEH statistic used in transport model validation (Department for Transport, 2026). It is a bounded, scale-independent quality metric with a score between 0 and 1, where 1 is a perfect match. It measures the error against a yardstick that grows with the square root of the count, so it tolerates a larger percentage error on quiet roads, where a few vehicles make a large percentage, and a larger absolute error on busy roads. It then maps the result to a bounded score. SQV is therefore consistent across the full range of traffic volumes. The formula is

SQV = 1 / (1 + sqrt((M − C)² / (f × C)))

where M is the modeled value, C is the observed count and f is a scaling factor set by the order of magnitude of the quantity: 1,000 for hourly volumes and 10,000 for daily volumes. An error of zero gives a score of 1; the larger the error relative to the count, the lower the score. For example, an hourly estimate of 1,100 vehicles against a count of 1,000 scores about 0.91. We report SQV as its 15th percentile across all segment-hours in each group: at least 85% of the segment-hours in the group perform better than the stated value. The quality thresholds below come from Friedrich et al. (2019):

4.5 Mean absolute error (MAE)

MAE measures the average absolute difference between predicted and observed volumes, in vehicles per hour. We report it for hourly volumes. Unlike the percentage-based metrics, it answers the question in traffic units: by how many vehicles per hour is a typical estimate off? An MAE of 30 means that hourly estimates differ from actual counts by 30 vehicles on average.

MAE complements MAPE (Section 4.1). Because MAE is measured in the same units as the traffic itself, the small denominators of quiet roads and quiet hours do not inflate it, although a difference of a few vehicles there registers as a large percentage error. The reverse also holds: high-volume roads dominate MAE, because the same relative error there means many more vehicles. MAE values are therefore most meaningful when compared within a volume category or road class, and read alongside the percentage-based metrics.

5. Validation approach: direct comparison with recent counts

The per-country metrics for 2026 show a direct comparison of the product with ground truth. We take the hourly volumes as the product delivered them for 2026 and compare them with the loop-detector counts recorded in 2026. Nothing is retrained for this comparison.

For past years, the results come from k-fold cross-validation and leave-one-country-out validation over the full set of loop counters (Sections 5 and 6 of the documents for those years). For the current year we test the product as customers receive it. The 2026 counts are the freshest ground truth available, and the recent ones are not part of the training data, so this comparison is an independent check of the delivered volumes against the loop counts closest to now.

Ground truth for the current year is still being collected, so fewer counters are available than for past years. The Counters column in the tables of Section 6 gives the number behind every figure; read a figure with few counters as an indication rather than a precise value. The counter data also covers a different period in each country; each country subsection in Section 6 states the period behind its tables.

The metrics are the ones defined in Section 4, computed against every counter with 2026 counts in each country. We report results for three volume categories. The volume category is a different grouping from road class (Section 1):

• High-volume roads: 55,000 or more vehicles per day (AADT)
• Medium-volume roads: 5,000–54,999 vehicles per day
• Low-volume roads: fewer than 5,000 vehicles per day

For the hourly estimates, a counter’s volume category is set by its annual average daily traffic derived from its own counts, not by the volume of the hour being scored.

The hourly results are measured per segment-hour against the counter’s count for that hour. Section 6 reports separate results for day hours (6am–11pm) and night hours (11pm–6am).

This comparison needs counter data in the country, so it gives no pooled figure for markets without counters. For those markets, the leave-one-country-out (LOCO) results in the documents for past years remain the accuracy basis.

6. Results (2026)

The tables below give the hourly accuracy results for each of the five countries in the 2026 direct comparison of the product with recorded counts (Section 5). Section 4.4 defines the quality tiers for the SQV (15th Pct) values.

6.1 Belgium

The counter data behind the tables below covers 1 January to 20 September 2026.

Day hours (06:00–23:00)

Night hours (23:00–06:00)

Belgium’s day-hour estimates stay close to the counts, with a median percentage error between -2.4% and 2.5% across the three road categories and a MAPE of 7.4% on high-volume roads. Night hours carry a positive median error that grows from 3.6% on high-volume roads to 11.1% on low-volume roads, where the night MAPE of 36.4% corresponds to a MAE of only 10.3 vehicles per hour and the SQV stays in the Very good tier (0.92). Every road category in both periods reaches at least the Fair tier, and the counter data, recorded up to 20 September 2026, covers the longest period of all countries in this assessment.

6.2 Netherlands

The counter data behind the tables below covers 1 January to 13 September 2026.

Day hours (06:00–23:00)

Night hours (23:00–06:00)

The Netherlands follows the United Kingdom closely, with a day-hour MAPE of 5.8% on high-volume roads and 8.4% on medium-volume roads and a median percentage error of 1.4% and 1.6%. Night hours more than double the percentage errors, as elsewhere in the document; on high-volume roads the night MAPE of 14.4% comes with an SQV in the Fair tier (0.84), while every other road category stays in the Good or Very good tier. The low-volume night MAPE of 33.8% goes together with a MAE of 7.3 vehicles per hour and a median error of 0.0%, so the percentage figure reflects small denominators rather than a bias.

6.3 Norway

The counter data behind the tables below covers 1 January to 12 July 2026.

Day hours (06:00–23:00)

Night hours (23:00–06:00)

Norway has no counters on high-volume roads, so its tables cover medium- and low-volume roads only. The median percentage error is positive throughout: 9.2% on medium-volume roads and 14.1% on low-volume roads in day hours, rising to 18.2% and 28.6% at night, the largest positive bias of all countries in this assessment, and the low-volume night MAPE of 56.6% is the highest in the document. In absolute terms the night errors stay small, with a MAE of 17.5 and 6.7 vehicles per hour, which is why every SQV value still reaches the Fair tier or better.

6.4 United Kingdom

The counter data behind the tables below covers 1 January to 13 September 2026.

Day hours (06:00–23:00)

Night hours (23:00–06:00)

The United Kingdom records the lowest MAPE of all countries in this assessment on high- and medium-volume roads in both periods, from 4.6% on high-volume roads in day hours to 14.1% on medium-volume roads at night; on low-volume roads it sits mid-field, at 16.1% by day and 33.1% at night. The median percentage error stays within 1.5% of zero on high- and medium-volume roads in both day and night hours, and the 95th percentile APE on high-volume roads (11.8% in day hours) is the narrowest in the document. All SQV values lie in the Good or Very good tiers.

6.5 United States

The counter data behind the tables below covers 1 January to 6 September 2026.

Note: The counters available for the current year are concentrated in a few states, so the figures reflect a limited part of the United States road network rather than a nationwide sample; a wider geographic spread will come as more count data arrives.

Day hours (06:00–23:00)

Night hours (23:00–06:00)

The United States shows the smallest gap between day and night hours of all countries in this assessment on high- and medium-volume roads, with a MAPE of 8.7% and 10.2% in day hours against 11.1% and 15.7% at night. A mild positive median error of 3% to 5% runs through every road category in both periods. On low-volume roads the night MAPE of 26.3% is the lowest in the document, with a MAE of 7.0 vehicles per hour, and the SQV reaches the Very good tier (0.94).

7. What the results mean for your use case

7.1 For business decision-makers and analysts

Traffic volume data informs strategic decisions (Section 2). For most analytical applications, the question comes down to one thing: is the error range acceptable for the decision at hand?

A practical guide: a MAPE of 10% on a road carrying 2,000 vehicles in an hour means that estimates differ from the true figure by about 200 vehicles on average, and the 68th and 95th percentile columns in Section 6 show how wide the error gets on a single road. For retail site selection, insurance risk modeling or transport infrastructure planning, an error of this size is usually acceptable. For applications that need precise capacity calculations, such as junction design or traffic signal optimization, we recommend adding local count data to the volume estimates where it is available and practicable. For comparisons across time, such as a before-and-after study, Section 8 explains how to read a difference between two periods against the stated accuracy, and how model improvements reach past periods.

Low-volume roads (under 5,000 vehicles per day) tend to show higher MAPE values (Das & Tsapakis, 2020). On roads with very low volumes, and in quiet hours such as the night, a higher percentage error still means a small number of vehicles — the worked example in Section 4.1 (a MAPE of 15% on a road carrying 40 vehicles in an hour, 6 vehicles) shows the scale. For use cases that depend on individual low-volume rural roads, treat the estimates as indicative and validate them against available count data where precision matters.

7.2 For data scientists and transport modelers

The SQV 15th percentile is a conservative quality indicator (Section 4.4 explains how to read it). When you integrate TomTom Historical Traffic Volumes into a transport model, the SQV shows which road categories you can use with confidence and where extra validation against local counts is advisable.

For road categories with SQV values at or above 0.80 (Fair or better), the data is suitable for transport models and analytical workflows that need segment-level accuracy. For categories between 0.75 and 0.79 (Acceptable), use the data for aggregate and indicative purposes and validate individual segments against local counts. For categories below 0.75 (Insufficient), treat the data as indicative and apply extra quality filters or local calibration where precision is required.

Where the median percentage error (bias) of a category is close to zero, aggregate measures such as total vehicle kilometers traveled across a network, or the average hourly volume for a road class, are reliable even where individual segments carry errors. Where a category shows a negative median percentage error (a tendency to under-predict), account for it in applications where absolute volume totals matter.

8. Updates, versions and comparability

8.1 Estimates are updated

TomTom Historical Traffic Volumes is a modeled product. We improve the model continuously, and an improved model can recompute the periods we have already published. A figure for a past period can therefore improve after publication: the recomputed estimate reflects a better model, while the traffic that occurred is unchanged. Regenerating the history in this way keeps a series internally consistent, because every period in it comes from the same model.

The set of contributing data sources also changes over time, as sources are added and removed, and such a change can affect where hourly estimates are available (Section 3.5).

8.2 Comparing periods

Every estimate in this product is a measurement with a stated accuracy: the quality figures in Section 6 give it for each country, by road category and time of day. A comparison between two periods, in a before-and-after study, a year-over-year trend or network monitoring, is a comparison between two such measurements.

The two products behave differently over time. AADT and AAWHT aggregate a year of observations, so short-term variation in the probe data largely averages out, and they are the natural basis for year-over-year comparison. The hourly estimates are designed to absorb changes in the contributing fleet (Section 3.4). Where a large change in data sources still shifts their accuracy or coverage, up or down, the published quality figures show it.

Our guidance: read a difference between two periods against the accuracy figures published for both periods. When we regenerate the history, refresh both periods, so that they come from the same model.

We will extend this guidance as the product evolves.

9. Summary

This document reports the accuracy of the TomTom Historical Traffic Volumes hourly estimates for 2026, measured by comparing the product’s estimates directly with the loop counts recorded in 2026 in five countries.

Across the five countries, the United Kingdom performed best, with a day-hour MAPE of 4.6% on high-volume roads and 7.7% on medium-volume roads and no measurable bias; the Netherlands, Belgium and the United States follow with day-hour MAPE values under 11% on high- and medium-volume roads and SQV values in the Acceptable to Very good tiers throughout. Norway’s median percentage error is positive in every road category and period, from about 9% to 29%.

Two points frame these results. The comparison uses the production estimates as delivered, so the figures describe the product customers receive rather than a retrained model; the counter data was recorded up to 20 September 2026 and extracted on 23 September 2026, and the ground-truth set ranges from about 1,750 counters in Belgium to 10,450 in the Netherlands, already matching or exceeding the 2025 training set in every country except the United States. And the counter data covers a different period in each country, from about six and a half months in Norway to more than eight months in the other four; a country enters this page only once at least twelve weeks of counter data are available for the year, and its figures will settle further as more data arrives.

On low-volume roads, MAPE values tend to be higher in percentage terms; as Section 4.1 explains, the difference in vehicles on these roads is typically small.

We publish the results for every validated country, including where they fall short of the quality thresholds in Section 4.4. With this document, customers have what they need to use TomTom Historical Traffic Volumes with a clear view of both its accuracy and its limits.

10. References

Das, S., & Tsapakis, I. (2020). Interpretable machine learning approach in estimating traffic volume on low-volume roadways. International Journal of Transportation Science and Technology, 9(1), 76–88. https://doi.org/10.1016/j.ijtst.2019.09.004

Department for Transport. (2026). TAG Unit M3.1: Highway Assignment Modelling (May 2026). Transport Analysis Guidance. https://www.gov.uk/government/publications/webtag-tag-unit-m3-1-highway-assignment-modelling

Eisinga, K., & Lorkowski, S. (2025). Network-Wide Traffic Volume Estimation Based on Probe Vehicle Data. Transportation Research Record, 2679(4), 264–277. https://doi.org/10.1177/03611981241289408

Federal Highway Administration. (2022). Traffic Monitoring Guide (Version 1.0, December 2022). U.S. Department of Transportation. https://www.fhwa.dot.gov/policyinformation/tmguide/tmg_2022/

Friedrich, M., Pestel, E., Schiller, C., & Simon, R. (2019). Scalable GEH: A Quality Measure for Comparing Observed and Modeled Single Values in a Travel Demand Model Validation. Transportation Research Record, 2673(4), 722–732. https://doi.org/10.1177/0361198119838849

Greenshields, B. D. (1935). A study of traffic capacity. Highway Research Board Proceedings, 14, 448–477. https://onlinepubs.trb.org/Onlinepubs/hrbproceedings/14/14P1-023.pdf

Herrera, J. C., Work, D. B., Herring, R., Ban, X., Jacobson, Q., & Bayen, A. M. (2010). Evaluation of traffic data obtained via GPS-enabled mobile phones: The Mobile Century field experiment. Transportation Research Part C: Emerging Technologies, 18(4), 568–583. https://doi.org/10.1016/j.trc.2009.10.006

Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679–688. https://doi.org/10.1016/j.ijforecast.2006.03.001

Kerner, B. S. (2004). The Physics of Traffic. Springer. https://doi.org/10.1007/978-3-540-40986-1

Sekuła, P., Marković, N., Vander Laan, Z., & Farokhi Sadabadi, K. (2018). Estimating historical hourly traffic volumes via machine learning and vehicle probe data: A Maryland case study. Transportation Research Part C: Emerging Technologies, 97, 147–158. https://doi.org/10.1016/j.trc.2018.10.012

Transportation Research Board. (2022). Highway Capacity Manual 7th Edition: A Guide for Multimodal Mobility Analysis. National Academies Press. https://doi.org/10.17226/26432

Treiber, M., & Kesting, A. (2013). Traffic Flow Dynamics: Data, Models and Simulation. Springer. https://doi.org/10.1007/978-3-642-32460-4

Zhan, X., Zheng, Y., Yi, X., & Ukkusuri, S. V. (2017). Citywide Traffic Volume Estimation Using Trajectory Data. IEEE Transactions on Knowledge and Data Engineering, 29(2), 272–285. https://doi.org/10.1109/TKDE.2016.2621104