AI × City Climate Action Hackathon 2026
Decide where to validate electric jeepney pilots first.
TL;DR: Screen 1,522 historic corridor records into 9 priorities for field validation, then show what the city must verify before spending.
Validation priority
9routes stay top-10 across policy tests
ScreenValidatePilot
Models organize evidence · people approve every real-world stepValidation-ready, not deployment-approved.20 dated map records · 0 field-confirmed active routes · 8 corridors prioritized for field validation.
Corridor Map
Compare routes, then open only the evidence you need
Reviewed OSM matches use observed member-way geometry. Other routes remain clearly labelled as historic screening records with planning geometry.
Use the route selector as the keyboard-accessible alternative to clicking map points. Dashed route lines indicate approximate or unverified source geometry.
Preparing the selected corridor…
0 routes shown Dashed = source geometry unverifiedMap © Mapbox · Data © OpenStreetMap
Route Lens
One corridor, one clear decision
Loading route…
Evidence detail Open the eight supporting signals+
PriorityDERIVED
—Live policy score
EvidenceDERIVED
—Confidence grade
ClimateSCENARIO
—Net tCO₂e / year range
EquityPROXY
—Population exposure only
ChargingPROXY
—Mapped context; capacity unverified
OperatorNEUTRAL PRIOR
—Awaiting consent-based evidence
RobustnessDERIVED
—Top-10 probability
TypologyML_ESTIMATED
—Comparable corridor pattern
What changes this decision?
Loading evidence sensitivity…
Current rank —Median —P10–P90 —
Feasibility snapshot
Put an order of magnitude beside the shortlist
Vehicle and charger figures are screening proxies, not supplier quotes or a budget. Financing remains explicitly missing.
Model restraint check
ML fills one missing historic activity field—nothing more.
Loading model comparison…
Scenario Lab
Compare policy choices without hiding trade-offs
Live ranks respond to normalized human-controlled weights. Robustness remains a labelled precomputed reference around the default lens.
Active policy mix
Compared with default
Phase-1 Portfolio
Build an evidence-validation shortlist, not a shopping list
The default eight-corridor result is precomputed by the deterministic pipeline. Changed constraints produce a clearly labelled interactive preview.
Selected corridors
Precomputed defaultWhy this differs from top-N
Active constraints
Corridor action queue
Top routes under the active policy lens
| Rank | Route | City | Priority | Evidence | Climate range | Robustness |
|---|
Evidence queue
Validate what could reverse the decision
Priority is based on deterministic field perturbation, not language-model confidence.
Planning & Evidence Assistant
Ask from the active route and scenario
The assistant receives structured evidence only. It cannot edit scores, policy weights or portfolio constraints.
Method & Sources
Transparent enough to challenge.
Open only what you need. The dashboard stays focused while assumptions, provenance and limitations remain one click away.
01 How to use Route2Zero+
- Choose the Metro Manila or LGU scope.
- Set a policy lens and inspect the live scenario ID.
- Select a corridor on the map or action queue.
- Read its eight signals and the evidence that could change the decision.
- Build a constrained Phase-1 validation portfolio and export the audit pack.
02 Why this decision matters+
Electrification resources and field-validation capacity are limited. Route2Zero helps city teams identify which corridors merit deeper work while keeping climate ranges, weak evidence and possible reversals visible.
03 Claims, assumptions and safeguards+
ML service intelligencePredicts historic service activity for anomaly detection and gap analysis; it is not ridership and is not used when the direct historic proxy is available.
Climate scenariosLow/base/high net CO₂e and energy results are deterministic scenario bounds, not statistical confidence intervals or measured reductions.
EquityUses population exposure only. No validated informal-settlement, socioeconomic or accessibility-gap layer is claimed.
ChargingMapped proximity is contextual evidence, not proof of utility capacity, interconnection or site control.
OperatorThe 50/100 value is a constant neutral prior across all routes, not evidence of operator readiness.
TypologyK-means silhouette is 0.373. The groups are descriptive, overlap materially, and contribute no policy points.
GovernanceML estimates. Deterministic models quantify. Policy weights remain human-controlled. The LLM explains and triages evidence; it never silently edits scores or choices.
04 Sources, models and pipeline health+
Loading build manifest…
05 Field observation intake+
Authorized validators can record dated route observations in the same schema used by the pipeline. Do not include rider names, phone numbers, or other personal data.