Use case

Member vs Casual Riders (Manhattan Demo) - Powered by Etherdata's Canonical US Census at H3 Resolution 8

EtherData.ai makes spatial data trustworthy and usable for every team. We publish open, rigorous, deeply usable datasets that make decision-making fairer and faster.

This demo answers a practical operator + city question: where is membership adoption strong vs weak, where is casual usage disproportionately high, and how those patterns relate to neighborhood socio-economics.

Focus
Membership adoption and casual usage mix. Equity-aware neighborhood segmentation using canonical census context.
Geography
Manhattan, New York (NY County), H3 Resolution 8.
Spatial unit
Canonical H3 grid (R8) with deterministic joins between trips and census context.
Outputs
Member penetration, casual-to-member ratio, trip frequency proxy, socio-economic context for segmentation.

Why census belongs in “Member vs Casual” analysis

Trip logs alone mostly reflect where stations exist and where people pass through (CBD, tourism, parks). They do not explain whether membership adoption is structurally high or low relative to neighborhood context. Etherdata’s canonical census layer provides stable signals that make adoption patterns interpretable and comparable across the city.

Datasets

Citi Bike trips (BigQuery public)

Trip events with rider type (subscriber vs customer) and station coordinates aggregated to H3 R8.

Etherdata canonical census (H3 R8)

Income and education attributes keyed to the same H3 cells for consistent context.

Methodology

Define the analysis grid

Retrieve Manhattan geometry and polyfill to H3 R8 to create the authoritative cell list.

Assign trips to H3

Convert start-station longitude/latitude to H3 indices and keep only cells within Manhattan.

Split trips by rider type

Aggregate trips into subscriber vs customer counts per H3 cell.

Attach census context

Join canonical income and education fields to the same H3 cells for socio-economic interpretation.

Compute adoption metrics

Calculate member penetration, casual-to-member ratio, and trip frequency proxies (trips per distinct bike_id by rider type).

Results & Interpretation (What the charts show)

Member Penetration vs Median Income

The scatter suggests that membership penetration is not tightly explained by income in this sample. Member penetration stays relatively high across a wide range of median income values, which implies the system has reached broad adoption where service exists.

This is a strong product story: the network appears to be used by members across diverse neighborhoods.
Outliers (lower penetration at similar income levels) are the most valuable: they suggest neighborhood-specific constraints (station density, safety/bike lanes, last-mile connectivity, or localized visitor dynamics).

Member Penetration vs Bachelor’s Attainment

This chart shows a clear socio-economic structure: education attainment rises with income and forms a strong gradient. It is not an adoption chart by itself, but it establishes the “context surface” that Etherdata provides. In practice, it enables consistent segmentation: you can compare adoption metrics against stable neighborhood context in a deterministic H3 framework.

The strong gradient validates that census rollups are coherent at H3 R8 and behave as expected.
This context is what makes adoption interpretation portable: the same pattern generalizes to other cities without redoing tract joins.

Member Penetration Map

Spatial surface of member penetration rate by H3 cell. This shows where recurring membership behavior is strongest versus weaker cells.

Use this map to spot pockets with structurally lower membership conversion.
Interpret together with trip volume and neighborhood context to avoid low-volume artifacts.

Casual-to-Member Ratio Map

Spatial distribution of casual-to-member ratio. Higher values indicate corridors with stronger casual skew relative to subscriptions.

Useful for pricing and conversion strategy targeting by geography.
Validate extreme ratios with denominator size (`trips_member`) in low-volume cells.

What we can conclude (and what we cannot)

Conclusions supported by the demo

Membership adoption is broadly strong where the system is active; income alone does not appear to dominate penetration.
Casual-heavy corridors exist and show much more extreme variation than member penetration, consistent with destination demand.
Etherdata census context materially increases interpretability: it allows you to distinguish “casual because destination” from “low membership because potential barriers,” using stable neighborhood signals.
Segmentation is actionable: it provides a short list of cell types to investigate with targeted interventions.

What we cannot conclude from these data alone

No causality claims (income/education do not “cause” membership) without controls for station density, bike lanes, safety, land use, tourism intensity, pricing, and events.
No true trip frequency per user because public Citi Bike data lacks a stable rider ID; bike_id is a proxy.
Very high casual-to-member ratios can be denominator-driven in low-volume cells; always interpret alongside trip volume.

Why Etherdata’s canonical census at H3 R8 is the enabler

This demo is intentionally simple from a modeling standpoint because the product being demonstrated is the data layer: a canonical, deterministic, H3-keyed census surface that can be joined to operational mobility data in minutes.

The value is operational:

Deterministic joins (same H3 key across datasets)
Stable socio-economic context for segmentation (income + education)
Auditability (metrics trace to known census fields)
Portability (same query pattern works for other cities and systems)

Overall summary

Member vs casual is not just an operational split—it is a lens on adoption, pricing leverage, and destination demand. With Etherdata’s canonical H3 census layer, you can interpret that split in neighborhood context and isolate the pockets where membership may be underperforming relative to local socio-economic structure.