# Dataset coverage — one flagship (all capabilities) + partial-coverage demos

The showcase strategy: **one comprehensive network exercising every visualization capability**, plus
several networks that each light up a subset — so a reviewer sees the full range without one giant
dataset hiding the individual features.

> **What ships publicly:** **ITS I-95 (VA)** is the published flagship data hub (network + INRIX/RITIS
> speed + VDOT sensors + probe + OD), cleared as public USDOT ITS demonstration data. Chicago Sketch and
> ARC Atlanta are the other live portal demos. Only the multi-GB I-95 *raw* source stays out of the repo.

| dataset | network | MOE (vol/VC/queue) | time-dep MOE | QVDF/TMC speed | paths / bundle | trajectories | space-time contour | sensors | OD | 3D/KML export |
|---|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|:-:|
| **ITS I-95 (VA)** ★ FLAGSHIP (published) | ✓ | ✓ | ✓ | ✓ (TMC) | – | ✓ | ✓ | ✓ | ✓ | ✓ |
| **Chicago Sketch** (live portal demo) | ✓ | ✓ | ✓ | (BPR) | ✓ | ✓ (+auto-synth) | ✓ | – | – | ✓ |
| ARC Atlanta (145k) | ✓ | ✓ | – | – | – | – | – (static profile) | – | – | ✓ |
| NVTA I-395 (TMC, local) | ✓ | ✓ | ✓ | ✓ (QVDF+INRIX) | – | (semi-dyn) | ✓ real breakdown | – | – | ✓ |
| Tucson (I-10) | ✓ | ✓ | – | – | – | – | – (static profile) | – | – | ✓ |
| West Jordan (UT) | ✓ | ✓ | – | – | ✓ (routes) | – | – | – | ✓ | ✓ |
| Sioux Falls | ✓ | – | – | – | – | – | – | – | ✓ | – |
| toy bottleneck/merge/signal | ✓ | ✓ | ✓ | – | – | – | ✓ (teaching) | – | – | – |

**ITS I-95 (VA)** is the **published flagship** for **digital-infrastructure demonstration**: one corridor
fusing GMNS network + INRIX/RITIS TMC speed + VDOT loop sensors + probe trajectories + GPS waypoints + probe
OD — the "connect-from-the-DataHub" data hub (`adapters/its_datahub.py`, `its_datahub.html`). Data source:
the USDOT JPO CodeHub **Data Cleaning and Fusion Tool**
(https://github.com/usdot-jpo-codehub/data-cleaning-and-fusion-tool), cleared as **public ITS demonstration
data**. The extracted `datasets/08_public_ITS_VA_1-95_sample/` ships; only the multi-GB raw source is
git-ignored. Live: [data hub](https://asu-trans-ai-lab.github.io/gui4gmns/portal_demo/i95/datahub.html) ·
[Kepler.gl](https://kepler.gl/demo?mapUrl=https://asu-trans-ai-lab.github.io/gui4gmns/portal_demo/i95/map.kepler.json)
· [deck.gl](https://asu-trans-ai-lab.github.io/gui4gmns/portal_demo/i95/deckgl/) ·
[Google Earth](https://asu-trans-ai-lab.github.io/gui4gmns/portal_demo/i95/gmns.kml).

**NVTA I-395** uses the **TMC-converted GMNS corridor network** (`cases/05_nvta_nb_am/network/`) — small,
corridor-only — for the real congested space-time speed contour (green free-flow -> red AM breakdown).
Kept local (INRIX-derived speeds are restricted); rendered for review, not committed.

**I-210E corridor state-views** (`docs/dashboards/I210E_*.html`) are gui4gmns samples that demonstrate
the corridor time-series capability: **calendar** (whole year by day/year), **network** (corridor map by
mean speed), **space-time** (milepost×time speed heatmap). Corridor speeds derive from **Caltrans PeMS**
(public agency data), packaged in the released **TrafficFlowBench-CA** benchmark. **Cleared for public
release** — attributed in each page footer and the gallery corridor section. They are the template for the
"ask the AI for more" capability: the same three views can be generated for other corridors (e.g. ITS I-95)
or other years from a GMNS network + a link speed / `link_performance` table (or an AI solver's output).