A Proposal for the Cycling Infrastructure Improvements Along the IBX, Based On STRAVA User Data

A proposal for the Cycling Infrastructure Improvements Along the IBX, based on STRAVA User Data

Authors: Elif Ensari Sucuoglu, Joao Paulouro and Eric Goldwyn, Published in: February, 2026

As part of the 2025 cohort of the Strava Metro for Academic Researchers Program, our team at the Marron Institute of NYU explored Strava Metro cycling activity data around the Interborough Express (IBX) corridor. This report synthesizes those findings into recommendations for improving the cycling infrastructure to support safe, convenient access to proposed IBX stations and, ultimately, future IBX ridership. For more information on the IBX project, please see our previous research and posts.

Walkability, and to a lesser extent, cycling infrastructure around transit stations has been extensively studied within Transit-Oriented Development (TOD) research and is widely recognized as critical to supporting demand along transit corridors. Walking and cycling are active modes of transportation that provide access to jobs, education, health care, food, recreation, and other resources in urban centers, and they also complement public transit by enabling first- and last-mile access and transfers between lines. The quality of the built environment plays a central role in determining whether people choose to walk or cycle rather than rely on private cars, which are polluting, inefficient users of urban space and discourage active travel.

For both modes, the proximity of destinations, the structure of the street network, and the diversity and intensity of street-level uses are among the most influential factors shaping travel behavior. For cycling, the availability of bicycles, bike lanes, and the quality and continuity of those facilities are particularly important, while for walking, factors related to safety, comfort, and street-level attractiveness strongly influence whether people choose to walk to their destination.

We therefore consider improvements to walking and cycling conditions around IBX station areas to be essential to the success of the project, as public transit, walking, and cycling function as an integrated system. People are more likely to walk in a city like New York where public transit options are abundant, while walkable streets make it easier and more attractive to reach nearby transit stops. Similarly, the availability of safe bike lanes, bike-share options, and secure bike parking can significantly expand the effective catchment area of transit stations, making it feasible to access stations or transfer between lines even when distances exceed a comfortable walk. With this objective in mind, we have explored the walking and cycling data provided to us by STRAVA Metro and analyze existing cycling activity around proposed IBX stations to identify opportunities for targeted infrastructure improvements that can expand access to the corridor.

Strava is a mobile application used to track walking, running, cycling, and other physical activity, with more than 180 million users across 185+ countries worldwide. Using mobile phones or wearable devices such as smartwatches, users record location, distance, duration, and other activity-related attributes. For the purposes of this study, the most relevant component of the dataset is the anonymized, aggregated location-based activity counts, which Strava provides to academic researchers through the Strava Metro for Academic Researchers Program.

At the same time, it is well established that Strava data is not representative of the general population. Strava users constitute a small subset of overall cyclists and pedestrians and are more likely to be male, within the age group 25-44 and according to our findings, employed in the knowledge-sector and less likely to own cars. One multi-city study comparing Strava counts with city-recorded cycling volumes found that Strava activity represented between 1 in 42 and 1 in 271 actual cycling trips across five North American cities. A separate study that included both cycling and walking trips in Oslo found Strava to capture between 0.9% and 4.4% (1 in 111 to 1 in 23 ) of actual average monthly trips, depending on the month. A literature review found an average of 5% representativeness across studies that analyzed STRAVA data.

Because the relationship between Strava activity and total walking and cycling volumes varies substantially by geography and season, estimating total activity levels in New York City would require a city-specific calibration model, similar to those developed in studies mentioned above. Such models typically rely on multiple counters placed on streets with a wide range of activity levels and traffic conditions. While developing such a model is beyond the scope of this report, we do compare Strava cycling counts with bicycle counts collected by the New York City Department of Transportation (DOT) at locations where counters are available (Figure 1).

After excluding the top 25% of observations and street segments with fewer than five daily Strava counts, we find that Strava data represents between 1 in 38 and 1 in 322 cycling trips and 1 in 35 to 1 in 1513 pedestrian trips recorded by DOT counters (Figures 2 & 3). The substantially lower representation of walking activity likely reflects the fact that walking is far more ubiquitous than cycling and therefore less likely to be recorded. Accordingly, we only present an analysis based on the cycling data and recommendations pertaining to the cycling infrastructure. See the Appendix Section for details on our aggregation and comparison methodology.

Keeping these caveats in mind, we do not treat Strava counts as estimates of total cycling volumes. Instead, we use the data to identify relative patterns of activity, such as commonly used routes, spatial concentrations of activity, and areas where people are more likely to cycle compared to others. This approach allows us to assess whether differences in infrastructure conditions and the built environment are associated with higher levels of observed activity, even if that activity reflects a limited and non-representative demographic. While future work may involve conducting additional counts on corridors without existing DOT counters, for the purposes of this study, we treat Strava data as a high-resolution, citywide signal of where recreational cycling activity already occurs and where connectivity improvements may be most impactful for both direct access to the IBX stations as well as connecting the corridor to the core cycling infrastructure of the city.

Figure 1: DOT Bike Counter Locations, Bike Lanes and IBX Station Isochrones

Strava Metro also provides disaggregated information by gender, age, and trip purpose. However, we found that these more detailed subsets exhibited lower correlations with DOT counts, likely due to smaller sample sizes. For this reason, we rely primarily on average daily activity counts, calculated by dividing annual counts by 366 for the year 2024.

Despite these limitations, Strava Metro’s datasets provide consistent, citywide cycling activity information at the street-segment level. While capturing only a fraction of total activity, it offers valuable insight into route choice, network gaps, and latent demand. Our findings further indicate that Strava users in New York City disproportionately choose to ride on streets with existing bike lanes or off-street cycling infrastructure. This suggests that cyclists in New York prefer to use bike lanes when available, and that cycling on streets without dedicated facilities often reflects a lack of viable alternatives. Accordingly, our recommendations for improving connectivity between IBX station areas and the existing cycling network are informed by both observed route use and activity occurring despite inadequate or missing infrastructure.

Figure 2: Average Daily Cycling Trips Recorded by DOT Counters and STRAVA Users

Figure 3: Average Daily Walking Trips Recorded by DOT Counters and STRAVA Users

Distribution of Cycling Activity Along the Corridor and Contributing Factors

We used STRAVA data in two ways. First, we aggregated cycling counts on Open Street Map (OSM) segments to street segments. Second, we aggregated cycling counts within each 10-minute isochrone- defined as the area reachable by a 10-minute walk from a station location in all directions- for all NYC subway stations and the proposed IBX locations. This approach allowed us to compare cycling activity across station areas, as well as associated demographic and built-environment characteristics known to influence travel behavior, such as built floor area ratio (FAR), retail FAR, and the size of the young adult population.

Isochrone Level Aggregation

In terms of STRAVA cycling activity, we found that the IBX station areas fall within the lower three quintiles when compared to all NYC subway stations (Figures 4 & 5). Stations in Manhattan receive substantially higher levels of cycling activity, as expected given higher population and employment densities, a greater concentration of amenities, and a more extensive cycling network relative to the rest of the city. For this reason, we also evaluate IBX stations relative to other stations within Brooklyn and Queens, rather than only citywide rankings (Table 1).

Among IBX station areas, the highest average daily cycling activity recorded by STRAVA users occurs at Roosevelt Avenue Station, followed by Myrtle Avenue and 4th Avenue Stations. In contrast, the station areas of Linden, Livonia, Remsen and Utica Avenue Stations fall within the lowest 10 percent of all NYC station areas in terms of cycling activity. When compared only to Brooklyn and Queens station areas, Linden Avenue Station remains among the ten lowest-activity stations, whereas no IBX station falls within the bottom ten citywide in terms of STRAVA-recorded cycling activity. Low STRAVA ridership around IBX stations is expected, also because the demographics in these areas skew against the STRAVA users in terms of age, income and vehicle ownership (Figures 6 & 7).

Figure 4: Daily STRAVA Cycling Trips Per Squre Kilometer by Isochrone

Figure 7: Daily STRAVA Cycling Trips Per Squre Kilometer by IBX Isochrone

Figure 5: Daily STRAVA Cycling Trips Per Squre Kilometer by IBX Isochrone

Figure 4: Households with a Yearly Income of $200k+, per Isochrone and IBX Station Isochrones, Figure 5: Households Owning 2 or more Vehicles per Isochrone and IBX Isochrones

Figures 6-7: Households with a Yearly Income of $200k+ and Households Owning 2 or more Vehicles per Isochrone and IBX Isochrones

Table 1: Cycling Trip Averages and Ranks Across New York City and Brooklyn-Queens

Station Name	Borough	Avg Bike Count/ km2/day	Avg Bike Count/km NYC rank (443)	Avg Bike Count/km Bq-Qn rank (254)
Roosevelt Avenue	Bk	34	228	107
Grand Avenue	Bk	13	304	162
Eliot Avenue	Bk	10	337	186
Metropolitan Avenue	Bk	12	320	172
Myrtle Avenue	Bk	32	231	109
Wilson Avenue	Q	16	277	149
Atlantic Avenue	Q	15	289	158
Sutter Avenue	Q	10	334	183
Livonia Avenue	Q	5	407	238
Linden Avenue	Q	4	420	247
Remsen Avenue	Q	5	414	244
Utica Avenue	Q	5	413	243
Flatbush-Nostrand Av.	Q	17	273	145
East 16th Street	Q	23	254	128
McDonald Avenue	Q	15	282	151
New Utrecht Avenue	Q	7	382	218
8th Avenue	Q	10	343	190
4th Avenue	Q	28	241	117
Brooklyn Army Terminal	Q	14	299	160

It is unsurprising that Roosevelt Avenue and 4th Avenue Station areas, both of which are among the three highest density IBX station areas, also record the highest cycling activity. These stations are well integrated into the city’s subway network and receive some of the highest daily subway trip volumes among IBX stations. The 4th Avenue Station directly intersects the 4th Avenue protected bike lane and lies close to the Shore Parkway Greenway, further supporting higher cycling activity. Myrtle Avenue Station’s proximity to the protected bike paths within Evergreens Cemetery, the dense shared-lane network of adjacent Bedford-Stuyvesant, and the neighborhood’s fine-grained, walkable street network likely explains the elevated cycling activity observed in this station area.

In contrast, the relatively low levels of cycling activity recorded around Livonia, Linden, Remsen, and Utica Avenue stations likely reflect a combination of factors, including limited bike lane availability, greater distance from the core cycling network, lower residential densities, and higher levels of private vehicle ownership.

Factors Affecting Cycling Intensity

To better contextualize the station-area patterns described above and inform the street-segment analysis that follows, we examined how cycling activity relates to a range of demographic, travel, and built-environment characteristics. Specifically, we analyzed correlations between STRAVA cycling volumes and subway ridership; demographic indicators such as income and household size; travel characteristics including average commute times; and streetscape and network attributes such as sidewalk widths and street connectivity. All variables were aggregated at the subway-station isochrone level for all stations across New York City.

Cycling intensity around subway stations is highly uneven, with a small number of station areas functioning as citywide cycling hubs, while most station areas fall within a substantially lower range of activity. Across New York’s 424 existing stations, the median station area typical street-segment intensity is approximately 39 daily trips per kilometer, with a 90th percentile value of 223 and maximum of 885. This is consistent with a highly concentrated pattern of hotspot station areas where cycling activity is structurally supported and culturally embedded. This distribution implies that “cycling-to-transit” is not simply a question of adding end-of-trip facilities. In the highest-intensity places, cycling is already embedded in a broader system of street design, network connectivity, and land-use patterns. Cycling and station ridership co-locate, but not in a simple linear way. Cycling intensity tends to be higher in station areas with higher ridership (Spearman ρ=0.61), but the relationship is not straightforward. Around many subway stations where ridership is high, STRAVA cycling activity is low. Only when high subway ridership is complemented with cycling infrastructure, street-network connectivity and favorable streetscape characteristics such as wide sidewalks, do we see significant increases in STRAVA ridership.

If we ignore the socio-economic features, bike-route coverage, street network and streetscape conditions constitute the clearest and most actionable correlates for cycling intensity in station catchment areas. Within subway station catchment areas cycling intensity correlates most strongly with:

Bike route coverage. Station areas with more extensive and higher-quality cycling infrastructure exhibit higher cycling intensity (Spearman ρ≈0.56-0.62).

Streetscape and walkability conditions. Wider sidewalks and related qualitative streetscape measures are associated with higher cycling intensity (avg sidewalk width Spearman ρ≈0.59).

Street-network connectivity and through-movement. Space Syntax measures (see Appendix) capturing movement potential and network connectivity show some of the strongest correlations with cycling intensity (intchoice Spearman ρ ≈ 0.66)

These built-environment relationships exist alongside strong socio-economic correlates, such as knowledge-sector employment, short to medium commute times, and lower auto-dependence. However, from a planning and implementation perspective, the bike infrastructure, streetscape quality and street-network represent the most directly actionable levers.

Cycling in the IBX Corridor

We began by examining simple pairwise relationships using Pearson and Spearman correlations. While useful as a first pass, these measures cannot capture the non-linear relationships and interactions that are common in urban systems. To address this, we also used a tree-based ensemble model that can learn more complex patterns in the data. We interpreted the model using SHAP (SHapley Additive exPlanations), which breaks each prediction down into the contribution of individual features, allowing us to see how different factors influence the outcome.

A SHAP summary plot (Fig. 8) was used to visualize feature importance while simultaneously indicating the direction and magnitude of each feature’s effect on the target variable across the full distribution of observations. We then compared the top and bottom ranked cycling intensity IBX station catchment areas, to the IBX and NYC medians, and plotted the interactions of the most important features in Fig. 9.

Figure 8: SHAP Summary Plot Showing Feature Importances

Figure 9: Interactions of the Most Important Features

Current (2024) cycling intensity in proposed IBX station areas is lower than the typical NYC subway station areas, but not uniformly low. Within the IBX, 4th Avenue and Roosevelt Avenue are at or above the observed NYC median cycling intensity baseline, while Myrtle Avenue and East 16th Street sit in the upper range of IBX observed baselines, albeit below the NYC median. As with the existing subway system more broadly, we do not expect cycling activity in the IBX corridor to scale linearly with station ridership in the absence of targeted on-street interventions.

Relative to the existing NYC station set, IBX station catchments show a distinct socio-economic and travel-structure profile, with implications for TOD and planning policy (Figures 6, 7 and 8). Compared with the 424 existing stations, IBX catchments currently show higher shares of very long commutes (60+ minutes), lower knowledge-sector employment share, and a more road-oriented travel structure. This points to a situation where IBX’s wider benefits are likely to be framed around accessibility gains and mode-shift potential, rather than simply intensifying existing cycling cultures. For cycling specifically, this supports a strategy that treats bike-to-transit as access design that includes street-network and safety programs, rather than as a post-hoc set of station amenities.

Consequently, we can classify IBX stations in three groups based on observed baseline cycling intensity:

Group A: main baseline stations (approx. NYC median). These should be prioritized for early cycling improvements to take advantage of existing activity and distribute the flow within the corridor, and include Roosevelt Avenue and 4th Avenue. The focus here is on high-quality bike-to-transit integration, protected approaches, junction safety, and secure parking.

Group B: mid-band baselines (NYC p25–median). These include Myrtle Avenue, East 16th Street, Brooklyn Army Terminal, Grand Avenue, Flatbush–Nostrand Av, Metropolitan Avenue, Atlantic Avenue, McDonald Avenue. The focus here should progress into barrier removal and cycle route network completion.

Group C: (below NYC p25). Additional attention needs to be given to IBX stations where we currently expect high ridership but low cycling intensity, such as Eliot, Wilson, Sutter, Livonia, Linden, Remsen, Utica, New Utrecht and 8th Avenue Stations, as these represent missed opportunities and a disconnect between active transportation in the street public realm and the transit network. The implication here is that cycling to and within the IBX will depend on creating safe east–west connections and continuous approach routes, not simply adding end-of-trip facilities at stations.

Street-Segment Level Ridership

Figure 11 presents cycling activity aggregated at the street-segment level across New York City with a focus on the IBX corridor. Class I and II bike lanes, as of 2024, representing protected and conventional (painted) bike lanes are shown in thick green and blue lines, respectively, while STRAVA cycling activity is shown in dark purple, purple and magenta, representing high-to-low daily counts ranging between 2,000 and 10 trips.

Our first observation is that the highest levels of cycling activity are concentrated in Manhattan and gradually decline toward the outer boroughs, largely mirroring the distribution of cycling infrastructure. Outside Manhattan, the most heavily used corridors include the Brooklyn Greenway and adjacent dense neighborhoods such as Greenpoint, Williamsburg, Brooklyn Heights, and Boerum Hill, likely due to bridge connections that have increasingly incorporated bike lanes over the past several decades.

Second, some of the highest cycling volumes occur within parks and parkways, including Prospect Park, Forest Park, and Ocean Parkway, all of which feature protected bike lanes physically separated from traffic or, in the case of Prospect Park, full roadway closures to motor vehicles (see Figure 10 for the existing Brooklyn Greenway and planned expansions). This pattern is also consistent with STRAVA’s tendency to capture a higher share of recreational cycling trips.

Third, cyclists overwhelmingly use bike lanes where they are available and rely on streets without infrastructure primarily when nearby alternatives do not exist.

Finally, several routes without dedicated infrastructure function as continuations of existing bike lanes and are used despite the lack of protection, particularly when they connect directly to other segments of the network or to major parks and greenways.

Figure 10: Greenway Expansion Plan from https://www.brooklyngreenway.org/

Most of IBX’s station areas remain distant from the city’s core cycling network and the greenways, with the exception of the southern terminal station at Brooklyn Army Terminal, which directly intersects the Brooklyn Greenway. Nevertheless, despite limited protected infrastructure and only a small number of conventional lanes extending eastward from dense neighborhoods such as Bushwick and Bedford-Stuyvesant, we observe substantial cycling activity along several routes that connect Manhattan and central Brooklyn to peripheral greenways and parks.

The 4th Avenue Station isochrone, adjacent to the Brooklyn Army Terminal isochrone, connects the 4th Avenue protected bike lane to the Brooklyn Greenway, while the Queens Boulevard protected bike lane passing through the Roosevelt Avenue Station isochrone provides a similar connection to the Central Queens Greenway. Accordingly, Roosevelt Avenue and 4th Avenue stations record the highest and third-highest cycling activity per kilometer of street segment among all IBX stations. Notably, Myrtle Avenue Station, despite lacking continuous protected infrastructure aside from a short segment along Cypress Hills Street, records the third-highest cycling activity per kilometer, driven largely by cycling along Myrtle Avenue itself and along the Fresh Pond Road–Cypress Hills Street–Central Avenue corridor, which connects the city’s central cycling network to the Evergreens Cemetery paths and the Queens Boulevard protected lanes.

Beyond these corridors, we also observe elevated cycling activity along Roosevelt, Grand, Eliot, Metropolitan, Pitkin, and Bedford Avenues. The first three directly intersect proposed IBX stations, while the latter corridors lie within one mile of multiple IBX stations. These routes appear to attract cyclists despite limited protection, likely due to a combination of their role as major arterials serving dense residential and employment areas and the connectivity they provide to recreational destinations and the broader cycling network.

Figure 11: Street-Segment-Level STRAVA Ridership

Cycling Infrastructure Potential Along the Corridor

New Bike Lanes

In cities where cycling mode share is low and private vehicle use is high, cyclists depend heavily on dedicated infrastructure to ride safely, making the presence of cycling infrastructure a primary determinant of route choice. As illustrated in our maps, New York City is no exception: cycling activity is highest where bike lanes exist, and especially where protected lanes are present.

Figure 12: Proposed Bike Lanes and Tiers

The greatest opportunity for expanding the cycling network along the IBX corridor lies in routes that already exhibit significant STRAVA ridership, extend outward from the existing network, and either close critical gaps or connect to parks with bike facilities, while also improving access to proposed IBX stations. These priority routes are shown in pink (1) in Figure 12. Secondary proposals, shown in dark purple, consist of short network extensions designed specifically to connect existing infrastructure to IBX stations. Third-tier proposals include longer corridors that also demonstrate existing ridership but would require larger capital investments.

Our first-tier proposals include:

1- Extending the protected bike lane on Broadway to Diversity Plaza by upgrading existing shared lanes and constructing new protected segments.

2- Installing a bike lane along Roosevelt Boulevard from Vincent Daniels Square to the existing 74th Street–Broadway stop on the 7 subway line.

3- Adding a bike lane along Grand Avenue between Metropolitan Avenue and 69th Street.

4- Upgrading and extending bike infrastructure along Myrtle Avenue between Willoughby Avenue and Forest Park Drive.

5- Installing a bike lane along Avenue H between Ocean Parkway and Ocean Avenue, with a direct connection to the proposed IBX station.

6- Closing the short gap in the 4th Avenue protected bike lane between 64th and 66th Streets.

Secondary proposals include:

1- Connections between the existing network and the 8th Avenue, New Utrecht Avenue, Livonia Avenue, Sutter Avenue and Wilson Avenue Stations.

Our third level proposals include:

1- A bike lane along Metropolitan Avenue between Scott Avenue and 74th Street.

2- A lane along Eliot Avenue between Metropolitan Avenue and 69th Street.

Ideally, this expansion would prioritize upgrading existing conventional lanes to protected facilities and constructing new protected lanes along corridors without any current infrastructure. However, recognizing budget constraints, wider arteries such as the Roosevelt Boulevard, Grand Avenue, Myrtle Avenue, Avenue H and 14th Avenue can be prioritized for protected lanes, while conventional lanes may be appropriate for shorter and narrower connector routes.

We do not propose additional cycling infrastructure connecting to the Atlantic Avenue Station, as this station already lies along Eastern Parkway, which is included in the New York City Greenway Expansion Plan (Figure 10) and would naturally enhance access to IBX stations as well as the broader cycling network.

Citibike Infrastructure

Figure 13: Avg. Daily Trips Originating from Citibike Docs in June 2024

We also examined the spatial distribution of Citi Bike stations across New York City and the average daily ridership originating from each station in June 2024. Both station locations and ridership patterns closely mirror those observed in STRAVA cycling activity, with the highest concentrations in dense neighborhoods and along the city’s core cycling infrastructure.

As the City seeks to expand the cycling network and increase ridership, extending Citi Bike coverage toward the outer boroughs and peripheral neighborhoods should be a complementary strategy. The proposed IBX is expected to catalyze new development and travel demand around its stations, presenting a timely opportunity to expand bike-share service in parallel with new transit investment.

One potential strategy is to prioritize expansion around IBX stations that already have strong subway access and/or relatively high levels of observed cycling activity, as well as their adjacent station areas, including Roosevelt Avenue; the Brooklyn Army Terminal–4th Avenue–8th Avenue corridor; and the Myrtle Avenue–Metropolitan Avenue station areas. Expanding Citi Bike in these locations would build on existing travel patterns while strengthening first- and last-mile connections to the IBX.

A second strategy is to target station areas that function as transit deserts, consistent with findings from our previous research, and use bike-share to improve access to IBX stations in these underserved locations. Eliot Avenue, Metropolitan Avenue, and Myrtle Avenue stations, as well as 8th Avenue, New Utrecht Avenue, Utica Avenue, and Remsen Avenue stations, are particularly well suited to this approach, as bike-share could provide a low-cost, flexible access option in areas with limited existing transit connectivity.

Conclusion

In this report, we present an analysis of cycling behavior recorded by STRAVA users in New York City, with a particular focus on opportunities for infrastructure improvements along the proposed IBX light-rail corridor. We acknowledge that the city’s cycling network would benefit from upgrades and expansions beyond the geographic scope of this study. However, examining the IBX corridor’s relationship to the existing and planned cycling network, including the city’s greenways, is especially valuable, both because new rail infrastructure and associated development can expand cycling demand, and because improved cycling access can increase IBX ridership by expanding station catchment areas.

We also recognize that STRAVA data does not represent all cyclists and disproportionately reflect male, young to middle aged, tech-savvy, recreational and higher-income riders. Nevertheless, STRAVA provides uniquely detailed street-level activity data, making it more spatially comprehensive than sources such as Census commute mode share data or DOT bicycle counts, which are limited to specific counter locations. This granularity allows for comparative analysis of cycling activity across all street segments in the city, even if the data represent only a subset of cyclists. Importantly, it also highlights corridors with substantial use despite limited or missing infrastructure, indicating clear opportunities for targeted investment.

Finally, while expanding and improving New York City’s cycling network would enhance safety, increase cycling activity, and potentially reduce vehicle miles traveled, we do not recommend funding these improvements through the IBX construction budget. Previous cost studies suggest that project betterments, including bike lanes, can introduce delays and cost pressures. Accordingly, we recommend that these proposals be evaluated and advanced as part of the city’s broader cycling network expansion program.

Appendix: Methodology

STRAVA Data Standardization and Center-line Generation

Strava activity data utilizes OpenStreetMap (OSM) basemaps that differ according to the data year. These basemaps appear to undergo minimal pre-processing by Strava, primarily involving the splitting of lines at intersections to create a network based on routing decision points. Consequently, all OSM edges, including non-standard tags and sidewalk edges, are often incorporated, presumably to enhance network completeness. In New York City, this results in inconsistencies: some streets feature multiple parallel edges and sidewalks — each potentially receiving distinct Strava counts presumably attributed to the nearest feature — while others are represented by a single centerline. This lack of standardization can skew analytical results and hinder the calculation of various metrics. Our primary objective, therefore, was to develop a unified centerline network that accurately reflects aggregated Strava counts.

The process begins by ingesting Strava activity data and OSM road network data from the corresponding basemap year. Both datasets are reprojected and standardized to New York City’s metric projection (EPSG:6538). A join operation on osm_id integrates essential OSM attributes. Initial exploratory data analysis informs our understanding of trip count distributions across different road types, leading to the exclusion of ‘crossing’ footways and the establishment of attribute-specific length thresholds for subsequent processing steps.

An important part of our aggregation methodology is the generation of a unified street centerline representation. This centerline incorporates aggregated Strava counts for “logical streets,” defined as street segments between intersections, which, consistent with Strava’s approach, are considered decision points. Each logical street segment is thus represented by a single centerline. The transformation involves standard geometric and data cleaning pre-processing. Topology simplification is performed by merging nearly collinear segments based on angular deviation. Street zones (corridors) are generated from vehicular centerlines to identify and associate relevant sidewalk segments. For extensive variables (e.g., trip counts), segments forming the same logical edge are merged using the maximum attribute value, while counts from distinct parallel edges within the street zone are summed onto the primary vehicular centerline. Intensive variables (e.g., average speed) are aggregated using a length-weighted average to mitigate the influence of very short segments. Finally, any remaining non-vehicular network elements, not associated with vehicular streets, are incorporated into the dataset. A visual inspection of the results shows how sidewalk inclusion in the original dataset overpowers and skews interpretation. We expect this to propagate downstream during data analysis if not handled with care. A simplified centerline aggregation minimises both these issues.

Figure A1: Trips per OSM Segment as Provided by STRAVA

Figure A2: STRAVA Trips Aggregated by Center-lines

STRAVA data aggregation at Isochrone Level

STRAVA data, aggregated at source to individual OpenStreetMap edges within logical streets, is used as a proxy for cycling activity. After calculating the length of each centerline, a spatial join with intersect operations was performed with 10-minute walking distance isochrones for each New York City subway station and the proposed IBX line stations, resulting in an aggregated geodataset including maximum, mean, median, sum, and count values (these last two specifically employed for subsequent density measurements). We then divided the sum of trip counts by surface area as well by total centerline length of each isochrone. This gave us a per-square-kilometer and per-meter intensities of cycling activity for all isochrones.

Analysis proceeded through both visual and statistical methods to compare cycling intensities with demographic and built environment data per isochrone. Choropleth mapping was used to identify spatial patterns of activity, while statistical analysis explored correlations and other relationships within the data.

Figure A3: STRAVA Avg. Trip Counts per Meter vs. Share of Knowledge Sector Workers Among Residents per Isochrone

Figures A3-A4: STRAVA Avg. Trip Counts per Meter vs. Share of Knowledge Sector and Trade Sector Workers within Residents of Isochrone

Figures A4-A5: STRAVA Avg. Trip Count per Meter vs. Ratio of Households Owning 3 or More Vehicles and Average Commute Distance in Miles

Figures A5-A6: STRAVA Avg. Trip Count per Meter vs. Ratio of Households Owning 3 or More Vehicles and Average Commute Distance in Miles

Space Syntax

Recognising that cities function as complex systems, we apply network analysis on the NYC subway network, and Space Syntax on the NYC street network. Space Syntax is a methodology initially developed at UCL by Hillier & Hanson for analyzing how street networks function as connected systems, and how their structure influences movement and accessibility. At its core it relies on two principal measures: Integration, which captures how close a street segment is to all others in the network, and is commonly associated with destination intensity (i.e. how likely a street segment will function as a destination), and Choice, which measures how often a segment lies on the shortest paths between all other segments (i.e. how likely a street segment will be passed through), and is widely found to correlate with movement flows and trade location. Two indicators show up as main features explaining the cycling intensity model: Choice and NACH cross-mode difference. The latter is based on normalized choice, and captures the difference between pedestrian and vehicular street network representations, highlighting locations where the network structure privileges one mode of travel over the other.

By Elif Ensari, Eric Goldwyn, Joao Neves|2026-02-10T01:21:32+00:00February 9, 2026|

Unlocking the IBX’ Transit Potential

Unlocking the IBX’s Transit Potential

Authors: Elif Ensari Sucuoglu, Joao Paulouro and Eric Goldwyn, Published in: August, 2024

The proposed Interborough Express is an opportunity to improve Brooklyn and Queens residents’ mobility and connectivity, the city’s most populous and largest boroughs, but also to transform the existing 14-mile corridor into a vibrant, livable and walkable urban landscape. The best way to maximize its potential is to implement comprehensive improvements that complement the existing neighborhoods, including upzoning, densification, mixed-use development, improved bus service and street redesigns. Such a holistic intervention would ideally begin before construction starts. These improvements will attract more residents and businesses to the corridor; thus, increasing ridership beyond current projections once it opens, saving thousands of travel minutes for local residents, reducing car-dependency, improving health and air quality as well as creating opportunities for new housing construction.

While the construction of the IBX can deliver all of these benefits, the MTA’s success will ultimately be measured by the access it creates and ridership it generates. But how can the city support IBX so that New Yorkers adopt it into their daily routines for traveling to work, school and other destinations? Transit ridership is correlated with population density, availability of jobs, mixed land use, existing transit service, and walkability conditions around stations. The busiest MTA subway stations are in neighborhoods with high residential and/or commercial densities, serve as important job destinations, connect to various modes of transit, and are walkable environments replete with amenities. We have quantified these characteristics around the proposed IBX stations, in an attempt to understand how supportive the current conditions are for generating ridership and how they can be improved.

We developed 10-minute isochrones, representing a 10-minute walk, around each of the 19 proposed IBX stations in order to collect data on residential and job density, land use, transit-service level and walkability. We also created isochrones and collected the same data for all of the 468 existing subway station complexes^[1] in New York City.

10-minute walking-distance isochrone map

We refer to Transit Oriented Development (TOD) research to evaluate how our quantitative analysis results compare to recommendations from city-agency reports and academic literature. TOD studies have found that certain levels of residential and employment densities, mixed land use, transit service quality and walkability is essential to attract sufficient ridership to justify the large capital expense. IBX current cost estimate is north of $5 billion. Cost estimates early in a transit project, prior to any significant design and engineering work have been completed, are unreliable; however, they are useful as we consider benefits versus costs.

Our maps reveal how areas within a walkable distance from the proposed stations compare to others around the city and which areas have the potential to be developed and improved to cultivate transit-friendly conditions. Below we present our analyses of residential density, job density and mixed land use, transit service level and walkability conditions along the corridor. We wrap up by offering strategic recommendations to the city on how to maximize the value of the investment in building the IBX.

Population and Built Area Density

The IBX corridor has the potential to establish a robust rider-base, especially if the low-density neighborhoods along it are upzoned to allow for more residential development. The existing research on transit-oriented development (TOD) suggests that high-residential densities support high transit usage. Queens and Brooklyn have much lower population densities than Manhattan.^[2] Along the IBX corridor, however, census tracts near the proposed Roosevelt Avenue, East 16th Street, 8th Avenue and 4th Avenue Stations have population densities comparable to bustling Manhattan neighborhoods like the Columbus Circle, Greenwich Village, Chelsea or Alphabet City.

Population density around the IBX stations range between 11,800 and 132,800 residents per square mile with a median of 47,657.^[3] This is almost double the density that has been found to ensure a 40% mode share for public transit in US transit districts (Renne and Ewing, 2013). Renne and Ewing (2013) found that districts with 10,000-25,000 residents per square mile saw transit ridership drop to 14%; and those with 4,000 and 10,000 residents per square mile, to 5%. The developable land within each IBX isochrone has an average population density above this 25,000 threshold. However, upon closer study, there are at least 20 census tracts in the corridor with residential densities below 25,000, with a land area adding up to 2.5 square miles- more than double the size of the Central Park- presenting an invaluable opportunity for additional development.

Population density map highlighting density below 25,000

Along the IBX corridor, residential unit density ranges from 3 to 72 units/acre , with a median density of 26 units/acre. Academic researchers and city agencies have adopted between 15-25 housing units per acre to be a target residential density in an urban neighborhood to support transit use^[4] (SWEEP 2024; VISION 2040, 2022; Renne and Ewing, 2013; MARTA p44, 2010; Cervero, 2007). IBX will likely attract substantial ridership from existing residents along the corridor. However, 69 census tracts out of 2199 that intersect with the IBX isochrones remain below a 20 housing units per acre threshold, concentrating around Grand, Eliot, Metropolitan and Utica Avenue stations. A closer inspection of these areas reveal many lots that are underutilized, which could easily generate multiple housing units with the construction of even low to mid-rise residential buildings.

Residential units density map, highlighting density below 15 units/acre

Metro Mall near the Metropolitan Avenue Station

Covert St and Irving Avenue near the Wilson Ave. Station

Some of these low-density lots contain large industrial facilities. But should they be preserved for the sake of the jobs they provide? Job density is widely considered to be one of the strongest predictors of transit ridership in the TOD literature; in one study, doubling of jobs near a transit station was found to increase transit commuting share by 73% for station areas (Renne and Ewing, 2013). So if these lots are indeed job-dense, more creative approaches than replacing these industrial buildings with residential ones should be considered, such as the development of more industrial mixed use buildings. In fact, as early as 2018, the Department of City Planning reported that 64% of the new jobs in transit-accessible light manufacturing districts of Brooklyn and Queens were in non-industrial sectors such as healthcare, retail, and information. So converting some of the corridor’s under-utilized industrial buildings into residential and office buildings would create more space for both housing and office needs. But do the lots that are low in residential density actually constitute high-density job centers?

Jobs and Mixed Land Use

With the exception of a few census tracts near the New Utrecht, Avenue I and Flatbush Avenue Stations, the job density along IBX is insufficient to support transit ridership. Only a quarter of the census tracts within the station isochrones have job densities higher than 20 jobs/acre, and only 15 have more jobs than 50 per acre; 20-50 being the range within which travel mode share shifts from single-occupancy vehicle use to transit use and walking (Frank and Pivo 2014). Precisely speaking the IBX corridor would need to add 222,485 new jobs to increase its overall job density to 35 jobs per acre, or the median of the range identified in the literature.

Residential density below 15 units/acre and employment density below 20 jobs/acre

The best way to introduce more housing and jobs along the IBX corridor would be to cultivate mixed-use. Mixed-use buildings sustain robust ridership throughout the day by attracting riders with different purposes and schedules, while also improving walkability by encouraging short, within-neighborhood trips (Ewing et al., 1994; Frank and Pivo, 1994). The dominant land use along the corridor is residential, with mostly single to two family, or multi-family walk-up buildings. These present an opportunity to densify the area by upzoning to allow 4-to-5-story^[5] mixed-use buildings, as well as by building larger-footprint buildings through land consolidation. We also found that the large, low-density and predominantly industrial census tracts with single or double-story warehouses often accompanied by large parking lots don’t actually provide significant job densities. Several such lots, highlighted in blue in our map, can be built up to accommodate many times as many square feet of commercial or mixed uses. We show the land use along the corridor and focusing on some of these low-density lots below.

Predominant land-use per census tract map (source: MapPluto 2023)

Lot-level Land Use

Low density lots: distribution centers (top), storage facility and sound studios (bottom)

One & two family lots near Flatbush-Nostrand Station

Low-density Industrial uses near Remsen and Linden Avenues

Low-density residential, auto-repair and other industrial uses near Utica Avenue Station

Looking at the combined density of population and employment reveals another critical aspect of the IBX’s right of way: it acts as a demarcation line between the high-density, mixed use areas to its north-west and the predominantly low-density residential neighborhoods to its south-east. This can partly be attributed to the right of way itself, acting as a physical barrier between its two sides. Some of this barrier effect can be minimized through design interventions, which should also be used to address walkability issues along the corridor to improve conditions around the stations. But another factor exacerbating the problem is that both subway and bus service declines on the south east of the line. The combination of dwindling transit and increased distance from Manhattan significantly increases travel time and accessibility to the largest job centers in Manhattan.

Residential and employment density

Transit Service

New transit lines benefit residents living within walking distance to their stations and also connect trips for passengers who can conveniently access them through other means of transit. In a densely populated urban context like New York City, renowned for its comprehensive public transit network, the subway and bus services have the potential to significantly expand the accessibility of a new line, extending its influence well beyond the walking distance of the stations. However, the geographic extent and frequency of service of the connecting transit lines, in other words, the convenience of access to the stations by public transit, has an impact on whether people will opt to make the extra trip. This is why TOD studies incorporate measures like public transport accessibility level (Kamurzzaman et al., 2014), node index (Bertolini, 1999; Reusser et al., 2008) or reachability (Zemp et al., 2011), which quantify the convenience of transit-service based on indicators like the diversity of transit options, frequency and reliability of service, number of directions served and number of stations within a given travel distance.

How much will the current transit network support the IBX by providing convenient connections to its stations? The IBX has already been planned to connect with 16 subway lines as well as the Long Island Rail Road. To determine the combined effect of the bus and subway lines along the corridor, we calculated the total number of weekday trips and unique routes that serve each station area based on the GTFS schedule data. As a reference for comparison, we visualized this information for every subway station complex in New York City.

Number of bus and subway trips that serve the 10-minute catchment area of subway stations

Among the newly proposed IBX stations, Flatbush Avenue (isochrone) receives 6,481 weekday bus trips, the highest among the IBX stations and Eliot Avenue receives 615 bus trips, the lowest among them all. Looking at subway service, Roosevelt Avenue (isochrone) receives 1,682 trips per weekday, the highest number of subway trips in the corridor, whereas Grand, Eliot, Utica, Remsen Avenues and the Brooklyn Army Terminal (isochrones) are not served by any subway lines at all. For comparison, the station isochrone served by the highest number of subway trips in all of New York is 42nd street – Bryant Park^[6] receiving 6,498 trips while Parsons Blvd Station in Jamaica Center receives 18,854, the highest number of bus trips per weekday.

Subway Stations	Weekday Bus Routes serving the 10-minute isochrone	Weekday Bus Trips serving the 10-minute isochrone
(IBX max) Flatbush Ave.	10	6,481
(IBX min) Eliot Ave.	6	615
(NYC max) Parson’s Blvd (Jamaica Ctr)	36	18,854
(NYC min) Roosevelt Island Main st.	2	340
	Weekday Subway Routes serving the 10-minute isochrone	Weekday Subway Trips serving the 10-minute isochrone	Weekday Rapid Rail Trips that will serve the 10-minute isochrone with the addition of IBX trips
(IBX max) Roosevelt Ave.	6	1,682	2,012
(IBX min) Grand, Eliot, Utica and Remsen Avenues, Brooklyn Army Terminal	0	0	330
(NYC max) 42nd street - Bryant Park	22	6,498
(NYC min) Roosevelt Island Main st.	1	114

Top and bottom ranking IBX isochrones for bus and subway service

The IBX will bring in 330 new weekday trips to all the station isochrones^[7], increasing Roosevelt Avenue’s rail trips by 20% and introducing 330 new trips to the stations that currently have no rapid rail service. This should boost ridership for all connecting lines to the IBX. But also, increasing bus service to the low-service stations like Grand, Eliot, Utica, Remsen Avenues and the Brooklyn Army Terminal stations would help support IBX ridership.

Walkability

IBX’s right of way cuts through several city blocks with dead ends where it meets most residential streets, and with trenches or overpasses at wider streets. The designs of the new station connections with the street network should reverse this interruption around 19 locations along the 14 mile corridor. Increased foot traffic will also enliven these areas and attract some commercial development near stations. However, without a well-planned set of urban design interventions to accompany densification and mixed-use development, many walking routes to stations will go through wide multi-lane traffic arteries, large parking lots surrounded by fences, blind facades enclosing industrial facilities, and frustrating detours in order to avoid dead ends, hitting but not providing access to the right-of-way.

IBX’s right of way cuts through city blocks with trenches or overpasses at wide streets.

Large parking lots, industrial facilities and dead ends where the streets meet the IBX

Walkability, both a part of TOD indicators and also studied independently, shares its primary indicators with TOD including density, mixed-use and transit-accessibility. Since the quantitative measures of TOD and walkability overlap , we did not conduct a walkability analysis of the IBX at the census-tract or block-group level. However, we did utilize Space Syntax, a street-network analysis method, and Convex and Solid-Voids, a 3d, street-level analysis method to evaluate walkability.

We plan to publish the details of these two analyses in separate posts. But our findings indicate that there are pockets of walkable neighborhoods around the IBX with well-enclosed, human-scaled, dense and mixed-use streets^[8] that are disconnected from each other. On the other hand, several large motor-ways cut through the corridor without bringing in the density that could enhance the enclosure and add mixed-use to activate these streets. With densification, introduction of mixed-uses and street redesigns, the walkability of the corridor should be improved to support the accessibility of the newly proposed transit stops and enhance ridership.

Below is an example from the Convex and Solid-Void measures: Height/Width of streets, which indicates the level of enclosure provided by buildings, walls, fences and other vertical elements bounding the street spaces. Where streets are wide, and surrounding buildings are low, as in the case of wide boulevards or highways without much density around them, this value drops. The pockets of more walkable neighborhoods with narrower streets and multiple-story buildings around them are seen in darker colors near Roosevelt Boulevard, Myrtle, Wilson, 8th and 4th Avenue Stations. Where we see low enclosure levels due to the widths of streets or low-rise development around them, lane-reductions, and addition of vertically bounding vegetation elements like tall trees with leafy canopies can be recommended to accompany densification strategies.

Street-Voids height/width measures

Construction of the IBX, if supported by land use changes, street-redesigns and strategies to attract residential and commercial development along the corridor, can connect 600,000 existing residents living within a 10-minute walking distance from its stations, and, hopefully, thousands of more new residents to job centers in Manhattan, Brooklyn and Queens. The City should also view building the IBX as an opportunity to encourage the construction of much needed housing, in a location already well connected to the city’s transit network and to provide access to many of Brooklyn and Queens’ underserved neighborhoods.

Conclusion

IBX is a promising project that will attract a strong rider base from Broolyn and Queens. However, even though its population density can be considered sufficient when compared to residential TODs in the literature, many of its stations, especially, Grand, Eliot, Metropolitan, Linden, Remsen and Utica Avenue Stations have low population densities and provide an opportunity to encourage development to match the multi-billion dollar investment that is being made. Development that will bring new employment opportunities to the corridor should also be encouraged. This is the perfect moment to bring mixed-use development to the corridor, which should ideally be initiated without waiting for the IBX to break ground, as a partial solution to the city’s housing crisis.

The building of the IBX is especially critical for the many transit-deprived areas along the corridor; we found that out of the 19 proposed stations’ isochrones, 15 received fewer bus trips than the median number of bus trips received by subway station isochrones, and 13 received fewer subway trips than the median number of subway trips received by subway station isochrones in the city. While IBX will bring 330 more transit trips to each isochrone, bus service should be improved around the stations to boost ridership.

Finally, walkability conditions should be considered along the corridor, so that small pockets of pedestrian-friendly neighborhoods are connected, and the reach of the stations are extended beyond their 10-minute isochrones through an attractive, safe and accessible street network.

[1] Transfer stations are considered single station complexes. I.e. Times Sq-42 St/Port Authority Bus Terminal, Lexington Avenue-59th Street, Herald Square-34th Street etc.

[2] Manhattan has about 72k residents per square mile whereas Brooklyn and Queens have 39k and 22k respectively.

[3] At the census tract level, ignoring the census tracts that significantly overlap with the existing right of way and others that are not developable.

[4] About 12,500-13,000 units per square mile.

[5] Department of City Planning’s City of Yes Housing plan includes a “Transit Oriented Development” proposal where up to 5 story apartment buildings can be built in R1-R5 zoned if the lots are within Greater Transit Zones, are large enough and on the short sides of blocks.

[6] This isochrone covers an area including Times Square, the Grand Central, the 42nd Street Port Authority, Herald Square 34th Street and 47th-50th Streets Rockefeller Center Stations.

[7] This is assuming 24 hr service with 5 min headway during morning (6-9am) and afternoon (4-8pm) peaks, 10 min headway during the day, and 20 min headway between 11pm and 6am.

[8] See Ewing and Handy, 2009 https://www.tandfonline.com/doi/abs/10.1080/13574800802451155

By Elif Ensari, Eric Goldwyn, Joao Neves|2025-03-13T14:33:52+00:00August 9, 2024|

The Interborough Express Study

Authors: Elif Ensari Sucuoglu, Joao Paulouro and Eric Goldwyn, Published in: May, 2024

In April, Governor Hochul unveiled her major transit infrastructure investments for fiscal year 2025. Among these projects, $52 million has been dedicated to the design and engineering of the Interborough Express, a new light-rail line set to link Brooklyn and Queens. IBX is expected to significantly improve transit access in Brooklyn and Queens, which have long been underserved by public transportation. Its greatest impact will be felt by the residents along the corridor, a majority of whom are low-income, transit-dependent, and from communities of color, thereby enhancing connectivity and promoting equity.

As we expand our research beyond the cost drivers of transit projects, we have decided to study IBX’s development potential, by promoting strategic land-use changes and street redesigns that we consider necessary to maximize the $5.5 billion investment that will be made to build the line. Even though IBX will increase job access for the residents living close to the proposed stations, the corridor is severely low density, underbuilt and predominantly residential. Moreover, many areas along the corridor are not walking-friendly, with wide arterial roads and inadequate pedestrian infrastructure. With our research, we seek to answer the questions of where the city should upzone, build more housing, encourage mixed-use development and carry out street redesigns. These changes would promote greater ridership on the line, reduce commute times, and ultimately expand transit-friendly areas in Queens and Brooklyn, reaching further east and south.

What is the IBX?

The Interborough Express is a proposed rapid transit project that will use an existing, 14-mile freight right-of-way in New York City. The line will run along the Bay Ridge Branch owned by MTA LIRR (southern 11 miles) and the Fremont Secondary owned by CSX (northern 3 miles), extending from Bay Ridge in Brooklyn to Jackson Heights in Queens.

A total of 19 stations have been proposed along the line that would provide transfers to 17 subway lines as well as the LIRR. The MTA predicts that the IBX will attract 115,000 daily weekday ridership and cost $5.5 billion. There are 900,000 New Yorkers and 260,000 jobs within a 0.5 mile of the corridor. The project is currently working its way through the engineering and environmental permitting process. To access the MTA resources and track progress on the project, go to this link.

The proposed Interborough Express Line. Source: https://new.mta.info/document/87606

Our Analyses

Our goal in sharing research incrementally and informally is to actively engage with the transportation community. We seek to discuss not only our findings and results but also the various data, methods, and tools we employ. It is our intention to share our data and code, as soon as we have time to clean and organize them according to sharable standards.

Below is a summary of the content we plan to share through these articles and some images to provide a sneak-peek into our upcoming articles.

The existing conditions regarding the demographics, availability of and access to jobs, street-network connectivity and travel behavior of current residents along the corridor. We use census data, tax lot data and the street network topology for these analyses and present them through choropleth mapping and accessible data visualization. The maps give a sense of how residential and commercial uses are distributed in the vicinity of the proposed stations, how population and job densities change across these areas, and the different building typologies and preferred commuting modes of the residents living along the corridor.

Travel behavior: Motor vehicle use along the IBX

A street-level study of the corridor, utilizing a 3-dimensional, morphological analysis method named Convex and Solid Voids. This method was developed at the University of Lisbon and both Joao Paulouro and myself have been involved in its development. I have utilized this method to measure walkability attributes in urban neighborhoods in my dissertation, which has informed part of the analysis we performed on the IBX. As walkability is an important criteria when assessing a transit corridor’s TOD potential, we decided to implement this street-level analysis to measure walkability-related attributes. One of our intentions was also to further improve the method itself.
- As part of the Convex and Solid-Void analysis study, we developed a workflow that identified boundary properties of each tax lot using Google Street View and planimetric data. We utilized image processing and ML algorithms to estimate whether a tax lot’s boundary existed, its height and visual characteristics. We used the results as input to calculate what we termed ‘cognitive height’. This is an important measure of Convex and Solid-Voids that represents the perceived height of a street’s volumetric surroundings, that is to say, the perceived heights of buildings and their immediate boundaries.

Convex and Solid-Void Analysis method 3d image

Convex and Solid-Void Analysis: 3d interface showing the IBX corridor’s Flatbush Avenue Station and stations to its north

We are working on estimating the ridership for the proposed IBX stations using two different methods simultaneously. Both methods have been utilized as alternatives to the traditional four-step model for predicting ridership for new transit lines. We are testing the STOPS software developed by the FTA, by feeding it our own data with the goal of comparing results to MTA’s initially stated 115,000 daily weekday riders. We are also developing our own ML algorithm based on census, tax lot, GTFS and network analysis data. We will share the results as well as the workflows for both methods, and anticipate discussing these comparatively.

While working on the IBX, we cultivated an interest in the Roosevelt Boulevard Subway, a proposed rapid transit line along a heavily trafficked artery in Philadelphia. Since most of our analyses are based on open data incorporated into a semi-automated workflow, we decided to implement them on the RBS and come up with TOD-oriented recommendations for the RBS corridor.

Land use map along the Roosevelt Boulevard Subway

Land use along the Roosevelt Boulevard Subway

We have also shared current conditions analysis of the RBS with a group of agency officials, advocates, journalists and NGO leaders in Philadelphia. The consensus among them was that ridership estimates will be the most important determinant to evaluate the feasibility of investment to build the RBS. In upcoming articles, we will discuss our predictions consecutively for the IBX and the RBS.

Locations and accessibility of jobs to residents through different modes of transportation as well as the potential impact of a new transit system on this, serve as important indicators for assessing the feasibility of investing in the system. To evaluate this, we utilized some existing transit modeling and network analyses tools, and identified geographies within and outside of the proposed transit corridors that will see the most significant changes in job accessibility. An article will focus on these findings.

map of NYC showing jobs and increase in their accessibility

Access to jobs by transit.

To stay updated on our interim findings, please follow our work on this page. Our final article will unveil our comprehensive report, complete with links to exclusive datasets, charts, maps, along with the measures, algorithms and code we’ve developed.

By Elif Ensari, Eric Goldwyn, Joao Neves|2025-03-13T14:32:48+00:00May 27, 2024|