Teaching with NYC Open Data: Publishing Student Civic Research Through Reproducible Workflows

Christian Martinez

Teaching with NYC Open Data

Christian A. Martinez

Brooklyn College, City University of New York (CUNY)

NYC Open Data Week 2026

Resources

Why This Matters

I realized my students were learning to β€œjuggle upside down.”

They weren’t struggling with the codeβ€”
they were struggling with the relevance.

The Shift

  • Replace generic datasets with NYC Open Data
  • Let students study their own neighborhoods, systems, and experiences

The Result

πŸ‘‰ Learning actually sticks

Today’s Format

Lightning Research Talks

Each student will briefly share:

  • Research question
  • NYC Open Data dataset used
  • Key analysis or visualization
  • Main finding
  • Why it matters

Student Presenters

Crystal Adote

NYC Leading Causes of Death and Environmental Complaints

Overview

My project examines the leading causes of death in NYC from 2007–2014, and indoor environmental complaints such as mold, indoor air quality, asbestos, and more from 2010 to the present. I wanted to explore these datasets and see whether there were any relationships between the two.

Data

Indoor Environmental Complaint Types

This is a Heatmap that conveys 5 of the leading causes of death over the years

Figure 1: This is a Heatmap that conveys 5 of the leading causes of death over the years

Leading Causes of Death

This stacked bar graph conveys the amount of indoor environmental complaints over the years

Figure 2: This stacked bar graph conveys the amount of indoor environmental complaints over the years

Data Cleaning and Analysis

  • Took out columns that were not needed
  • Kept only five common, well-known causes of death
  • Paired complaint types with causes of death
  • Merged the two datasets

Statistical Tests

  • Correlation: r = 0.6122905
    • Moderately strong
  • Linear regression suggested that indoor complaint types did not show a meaningful relationship with the leading causes of death in NYC.

Takeaway

  • This topic is important to the general community because it sheds light on indoor environmental hazards that individuals file complaints about.
  • It also raises questions about whether there may be a relationship between indoor environmental hazards and leading causes of death in NYC.

Jonah Dratfield

Social Infrastructure & Well-being

The Overarching Questions

  • Can publicly available data be used to explore the conditions that best facilitate social connectedness, and thereby, most enhance quality of life?

  • Is there already data that points to β€œabstract” psychological constructs like well-being, loneliness, etc?

  • If so, is how can this data be acted upon or improved?

Some (Partial) Answers

  • At present, NYC Open Data does not include the validated measures psychologists typically use to assess metrics like social connectedness and well-being.

  • However, there are various proxies.

  • Permitted events are a proxy for connectedness.

  • Number of SNAP Benefit Recipients is a (very) rough proxy for economic health (which is often associated with well-being).

Community District/SNAP Chart

Figure 3

Results

  • A linear regression was conducted to determine whether number of permitted events predicts number of SNAP recipients.

  • The model was statistically significant, F(1, 723) = 45.34, p < .001, and explained approximately 6% of the variance in SNAP recipients (RΒ² = .059).

  • The number of events was a significant negative predictor of SNAP recipients, b = βˆ’21.30, SE = 3.16, t(723) = βˆ’6.73, p < .001.

Permitted Events/SNAP Linear Model

Figure 4

Limitations

  • Results are significant and promising … but …

  • SNAP is an imperfect measure of holistic well-being (as well as economic). We need more β€œmiddle range” data.

  • We need better social gathering info (Reddit, Meetup, Eventbrite, etc.)

  • Community districts are imperfect units. Access is as important as location. Parks, for instance, were excluded from the analysis.

New Directions

  • Survey data about abstract constructs to corroborate and inform the β€œpractical” data.

  • Use of evidence to intervene in low-barrier ways (and tracking of those interventions).

Joyce Escatel-Flores

NYC Art Meets NYC Appetite

Goals of the Project

  • Explore whether restaurants near art museums are more likely to have higher ratings than restaurants not close to museums.

  • Explore whether restaurants near museums are less likely to have no violation citations than restaurants not close to museums.

  • Creating an interactive map that pinpoints restaurants that are nearby museums.

Data Sets Used

  • DOHMH New York City Restaurant Inspection results which you can find at https://data.cityofnewyork.us/Health/DOHMH-New-York-City-Restaurant-Inspection-Results/43nn-pn8j/about_data
  • MUSEUM which you can find at https://data.cityofnewyork.us/Recreation/MUSEUM/fn6f-htvy/about_data

The third data set is a Kaggle open data set created by Beridzeg45 called NYC Restaurants, which you can find at https://www.kaggle.com/datasets/beridzeg45/nyc-restaurants

Goal 1: Visualizing the Analysis

We are visualizing the proportion of rating groups (high, medium, low) by if restaurants are near museums (yes,no).

Figure 5

Chi-Square Analysis for Goal 1


    Pearson's Chi-squared test

data:  contingency_table
X-squared = 1.1849, df = 2, p-value = 0.553
Cramer V 
 0.04365

  • The chi-square test shows X^2 = 0.64691 (0.65), df = 2, p = 0.7236 (0.72)

  • There is not a statistically significant relationship between restaurants being near museums and rating.

  • Cramer’s V tells us that the relationship is weak in strength.

Goal 2: Visualizing the Analysis

We are visualizing the proportion of if a restaurant ever had violations (None, Critical) by if restaurants are near museums (yes,no).

Figure 6

Chi-Square Analysis for Goal 2


    Pearson's Chi-squared test with Yates' continuity correction

data:  contingency_table_2
X-squared = 1.4637e-28, df = 1, p-value = 1
                    Near_Museum
Restaurant_Violation        No       Yes
            Critical  30.38585  4.614148
            None     509.61415 77.385852
Cramer V 
0.007957

  • The chi-square test shows X^2 = 6.2237e-30, df = 2, p = 1

  • Cramer’s V tells us that the relationship is very weak in strength (0.008128).

Continuation


    Fisher's Exact Test for Count Data

data:  contingency_table_2
p-value = 0.7978
alternative hypothesis: true odds ratio is not equal to 1
95 percent confidence interval:
 0.3339362 3.0818476
sample estimates:
odds ratio 
 0.9060068
  • p value is 0.7975, which is still not a statistically significant relationship between restaurants being near museums and restaurant violations.

Goal 3: Creating an interactive Map

Figure 7

Key takeaway

  • Red: Low Rated Restaurants
  • Purple: Average Rated Restaurants Dark
  • Blue: High Rated Restaurants.
  • Black: Museums

Project Relevance to New Yorkers

I believe that this project is relevant to New Yorkers who like to go to museums or restaurants and would like to plan an outing for a nice museum day in NYC. These types of New Yorkers would care about this type of project because they no longer have to rely on using Google to search each individual museum and instead have a map that is accessible and easy to use.

Thank you!

Robert Hutto

The Madison Square Garden Effect in the NBA: Using basic statistics to investigate The Garden’s influence on basketball performances

Background

  • Oldest active NBA arena (est. February 11, 1968)
  • Home-court of the New York Knicks, one of the league’s most valuable franchises
  • NYC’s premier indoor venue comes with prestige and disproportionate national coverage
  • Dozens of celebrities, athletes, public figures in attendance of every game
  • Iconic performances: Jordan’s 55-pt return from retirement (1995), Kobe’s MSG record-setting 61-pt game (2009), Curry’s 54-pt 11/13 from three breakout game (2013)

Madison Square Garden, β€œThe Mecca of Basketball”

The narrative: MSG uniquely influences players’ performances under its bright lights.

But is that narrative supported by data?

Research Questions

  1. Q1: Do the Knicks experience a special home-court advantage at MSG compared to other NBA teams at their home arenas?

  2. Q2: Do visiting players perform differently at MSG than at other away arenas?

  3. Q3: Which players benefit most (or least) from playing at MSG?

Methods & Data

  • Source: hoopR R package: NBA game and player box score data, 2002–present
  • Scope: Standard regular season and playoff games only
  • Key metrics: Points, True Shooting % (TS%), Turnovers, Offensive Output (PTS + REB + AST), Defensive Output (STL + BLK)
  • Player analyses excludes Knicks players home games; MSG data reflects visiting teams only
  • Comparisons use independent samples t-tests

Q1: Knicks Home-Court Advantage

  • The Knicks rank 23rd out of 32 teams in home-court scoring advantage
  • Bottom 1/3 of the league.
  • MSG as a home-court may actually be more of a detriment than a benefit.

Q2: Visiting Player Performance at MSG

Figure 8

Key takeaway

  • Less defense at MSG
  • Higher true shooting
  • Fewer turnovers
  • Suggests players may be more locked in offensively at MSG

Q3: Who rises to the occasion at MSG? Who struggles?

Figure 9

The influence of playing at MSG on individual players’ statistical production depends on the player.

Knicks fans:

  • Do the players on the left seem overrated by other NBA fans? Any NBA villains?

Conclusions

  • Q1: The Knicks’ home-court advantage at MSG is actually worse than most NBA teams.
  • Q2: Visiting players show small but statistically significant differences at MSG compared to other away arenas. On average, players shoot more efficiently and turn the ball over less, but produce less blocks and steals.
  • Q3: There is meaningful individual variation in how players perform at MSG relative to their road averages.
  • Takeaway: While players across the NBA perform better offensively at MSG, the β€œMSG effect” is not a homogeneous influence on players; some perform better and others worse when they play at MSG compared to other away arenas.

Isley Jean-Pierre

Examining How Juvenile Probation And School Discharge Contribute To Recidivism

Introduction

  • Juvenile recidivism remains a major policy concern
  • Probation supervision and school outcomes may influence rearrest rates
  • This project investigates:
    • Rearrest Rates, Probation Caseloads, and School Discharge Patterns

Goal: To help policymakers evaluate whether current probation resources are sufficient to reduce recidivism among youth.

Research Questions

  1. How does supervision caseload relate to rearrest rates among youths?

  2. What can school discharges tell us about supervision caseloads and rearrest rates?

Results

How does supervision caseload relate to rearrest rates among youths?

  • A correlation analysis (r = 0.49) shows a weak positive relationship between supervision caseloads and rearrest rates among the youths.

  • A regression analysis (p < .003) indicates that juvenile caseloads significantly predict rearrest rates. R-squared = 0.24 (24%).

  • Overall, these results signify that more caseloads tend to lead to more rearrest rates.

Time Series Analysis

[1] 0.4935201

Figure 10

Regression Analysis

Figure 11

Results

What can school discharges tell us about supervision caseloads and rearrest rates?

  • An independent t-test (t = 1.15, df = 159.43, p < 0.25) shows no significant difference between school discharge rate and school level.

  • A Chi-square analysis (X-squared = 938.62, df = 1, p < 2.2e-16) suggests a significant difference between discharge category and school level.

  • Cramer’s V (0.54). Moderate to strong relationship between discharge category and school level.


    Welch Two Sample t-test

data:  discharge_rate by school_level
t = 1.1492, df = 159.43, p-value = 0.2522
alternative hypothesis: true difference in means between group High School and group Middle School is not equal to 0
95 percent confidence interval:
 -0.008913023  0.033721970
sample estimates:
  mean in group High School mean in group Middle School 
                 0.06976680                  0.05736233 

    Wilcoxon rank sum test with continuity correction

data:  discharge_rate by school_level
W = 3034, p-value = 0.381
alternative hypothesis: true location shift is not equal to 0

Discharge Rates By School Level

Figure 12

Discharge Types By School Level

Figure 13

School Discharges By Districts

Conclusion

  • Caseload size is significantly related to rearrest rates
  • School discharge patterns differ by school level
  • Too many caseloads may reduce supervision effectiveness
  • Education disruptions can potentially lead to crime involvement
  • Demographic variables

THANK YOU!

Shannon Joyce

Project Overview

  • Mold Exposure β€”> Psychological Stress/Aggression

  • Psychological Stress/Aggression β€”> Domestic Violence

Do domestic violence reports and residential mold complaints in NYC follow similar, correlated patterns over time?

Datasets: Mold Complaints

311 Service Requests

Table 1. Aggregated residential mold complaints by borough, 2010-2024.
Year Month Borough Mold Complaints
2010 01 - January BRONX 954
2010 01 - January BROOKLYN 779
2010 01 - January MANHATTAN 410
2010 01 - January QUEENS 315
2010 01 - January STATEN ISLAND 58

Datasets: Domestic Violence Reports

NYPD Complaint Data Historic

Table 2. Aggregated domestic violence reports by borough, 2010-2024.
Year Month Borough DV Reports
2010 01 - January BRONX 910
2010 01 - January BROOKLYN 1306
2010 01 - January MANHATTAN 541
2010 01 - January QUEENS 791
2010 01 - January STATEN ISLAND 154

Exploratory Analysis: Heat Maps

Figure 1. DV reports & mold complaints by borough and year, 2010-2024. Darker colors represent a higher volume of complaints/reports.

Correlation Plot: Totals by Borough

Figure 2. A scatterplot representing a positive correlation between total mold complaints and DV reports, grouped by borough.

Correlation Test: DV Reports ~ Mold Complaints (Month to Month)


    Pearson's product-moment correlation

data:  x and y
t = 5.1733, df = 178, p-value = 6.155e-07
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 0.2272817 0.4822876
sample estimates:
      cor 
0.3615268
  • r = 0.36
  • p < 0.001

Correlation Plot: Lagged Data

Figure 4. A scatterplot representing a positive correlation between current-month mold complaints and DV reports 3 months later.

Correlation Test: Lagged Data

Current Month Mold Complaints ~ DV Reports 3 Months Later

  • r = 0.25
  • p < 0.001

Regression Models

Linear Regression Model \(R^2\) p-value AIC
Mold ~ DV + Borough 0.94 < 0.001 10571.03
Mold ~ DV + Borough + Avg. Resolution Days 0.95 < 0.001 10524.65
Mold ~ DV (3 Months Later) + Borough 0.94 < 0.001 10390.65

Key Findings & Takeaways

  • Housing distress πŸ”— Domestic instability
  • Predictive power of both models
    • Response speed
    • 3-month lag
  • Proactive roadmap for NYC agencies
    • NYC Housing Authority
    • NYC Department of Health & Mental Hygiene

Full Project: https://rpubs.com/shannonjoyce/toxichomes

Emma Valentina Tupone

Environmental Stressors and Social Complaints in NYC

Project Overview

  • Climate change increases urban environmental stress

  • Flooding can disrupt infrastructure and communities

  • Social stress may appear in complaint behavior

Research Question

  • Do flooding complaints relate to noise complaints across NYC boroughs?

Data Preview

flooding

Table 1
Unique Key Created Date Closed Date Agency Agency Name Complaint Type Descriptor Location Type Incident Zip Incident Address Street Name Cross Street 1 Cross Street 2 Intersection Street 1 Intersection Street 2 Address Type City Landmark Facility Type Status Due Date Resolution Description Resolution Action Updated Date Community Board Borough X Coordinate (State Plane) Y Coordinate (State Plane) Park Facility Name Park Borough Vehicle Type Taxi Company Borough Taxi Pick Up Location Bridge Highway Name Bridge Highway Direction Road Ramp Bridge Highway Segment Latitude Longitude Location
67148990 2025-12-12 22:05:00 NA DEP Department of Environmental Protection Sewer Street Flooding (SJ) NA 10314 NA NA NA NA WILLOWBROOK ROAD FILLMORE AVENUE INTERSECTION STATEN ISLAND NA NA Open NA NA NA 02 STATEN ISLAND STATEN ISLAND 946947 159179 Unspecified STATEN ISLAND NA NA NA NA NA NA NA 40.60351 -74.13434 POINT (-74.1343370982284 40.60350738539052)
67150637 2025-12-12 18:14:00 NA DEP Department of Environmental Protection Sewer Street Flooding (SJ) NA 11385 NA NA NA NA STEPHEN STREET FOREST AVENUE INTERSECTION QUEENS NA NA Open NA NA NA 05 QUEENS QUEENS 1012116 194351 Unspecified QUEENS NA NA NA NA NA NA NA 40.70008 -73.89950 POINT (-73.8995024633561 40.70008149366179)
67150638 2025-12-12 14:50:00 NA DEP Department of Environmental Protection Sewer Street Flooding (SJ) NA 11236 713 EAST 86 STREET EAST 86 STREET GLENWOOD RD FLATLANDS AVE NA NA ADDRESS BROOKLYN NA NA Open NA NA NA 18 BROOKLYN BROOKLYN 1009167 172300 Unspecified BROOKLYN NA NA NA NA NA NA NA 40.63957 -73.91022 POINT (-73.91021940592279 40.6395652550322)

noise

Table 2
unique_key created_date complaint_type borough city x_coordinate_state_plane y_coordinate_state_plane
67857528 2026-02-05T10:35:00.000 Noise BRONX BRONX 1032409 257950
67897948 2026-02-08T08:17:00.000 Noise QUEENS OZONE PARK 1029496 189566
67822457 2026-02-02T15:08:00.000 Noise QUEENS JAMAICA 1045357 185786

Exploratory Summary

  • Flood complaints vary across boroughs and years

  • Noise complaints show even larger variation

  • Some boroughs report very high complaint activity

Flooding Complaints by Borough

Figure 14

  • Visual comparison of flooding complaints

  • Shows variation across boroughs

  • Highlights environmental vulnerability differences

Correlation

[1] -0.002238207

Figure 15

  • Each point = borough-year observation

  • Line shows overall trend

  • Relationship appears weak - 0.009

Regression Model

  • Tests predictive relationship

  • Coefficient β‰ˆ 0.42

  • p-value = 0.71

  • Not statistically significant

Key Takeaways

  • Flood complaints vary across NYC boroughs

  • Noise complaints vary widely

  • Relationship between the two was weak

  • Environmental stress likely influenced by multiple factors

  • NYC Open Data enables civic research

Thank You!

Questions?

Xinru Wang

Beating Around the Bush: Urban Trees and Wildlife Patterns in New York City

Urban Wildlife in NYC

  • Wildlife incidents are reported across NYC every day.

  • But they are not evenly distributed across boroughs.

  • What explains these patterns?

Could Street Trees Influence Wildlife Incidents?

At first glance, this might seem unlikely. Street trees line sidewalks, while many wildlife incidents occur in parks. But urban ecosystems are connected.

Street trees can support urban wildlife by providing:

β€’ Food
β€’ Shelter
β€’ Travel pathways

2015 NYC Street Tree Census

β€’ Over 680,000 street trees recorded across NYC
β€’ Includes species, location, and health condition
β€’ Used to estimate urban canopy coverage

Figure 16

Urban Park Ranger Wildlife Incident Reports

  • Reports of wildlife incidents across NYC parks
  • Includes injured animals, distressed wildlife, and conflicts with humans
  • Allows us to track wildlife activity patterns

Figure 17

Street Tree Density Across New York City

Figure 18: Brighter colors indicate higher concentrations of street trees.

Wildlife Incidents per 10,000 Street Trees

Figure 19: How many wildlife incidents occur for every 10,000 street trees.

Weak Relationship Between Street Trees and Wildlife Incidents

Figure 20: Each point represents a NYC borough.

Raccoons Dominate Wildlife Incidents Across NYC

Figure 21: Raccoons appear most frequently in wildlife incident reports across boroughs.

Key Takeaways

  • Street tree abundance alone does not strongly predict wildlife incidents

  • Wildlife incidents vary across boroughs

  • Raccoon are the most commonly reported species

  • Other urban factors likely drive wildlife encounters

Laura Werner

Domestic Violence Incidents vs Resource Allocation Across NYC

Research Question

  • Are domestic violence resources for victims meeting the needs of victims in New York City?

  • This project compares reported domestic violence incidents with Family Justice Center (FJC) service utilization.

  • The analysis focuses on 2020 so that incidents and service usage are directly comparable.

  • The goal is to determine whether boroughs with greater reported need also show stronger support service engagement.

Why This Matters

  • Domestic violence is a major public safety and public health issue.

  • Harm extends beyond immediate injury and includes long-term emotional, psychological, and developmental consequences.

  • Children exposed to violence in the home may also experience lasting effects.

  • Timely and effective support services are critical for survivor safety, recovery, and prevention of future harm.

Data Sources via NYC Open Data

Key Interpretation

  • The Bronx had the highest total number of reported domestic violence–related incidents.

  • Queens followed next.

  • Manhattan and Brooklyn showed similar moderate levels.

  • Staten Island had the fewest reported incidents.

Incident Type by Borough

Figure 23

Key Interpretation

  • Family domestic incident reports dominate across all boroughs.

  • Felony assaults and rape-related offenses occur at much lower frequencies.

  • The Bronx remains consistently high across most incident categories.

Family Justice Center Services by Borough

Figure 24

Key Interpretation

  • Family Justice Center client visits are much higher than services being provided.

  • Queens shows the highest overall service utilization.

  • Manhattan and Staten Island show lower totals across many categories.

Comparing Incidents and Client Visits

Figure 25

Key Interpretation

  • Boroughs with more reported incidents generally have more client visits.

  • However, the relationship is not proportional.

  • The Bronx has the highest incident burden but not the highest number of FJC client visits.

FJC Client Visits per 100 Reported Incidents

Figure 26

Key Interpretation

  • Standardizing visits by incident burden reveals sharper disparities.

  • Staten Island has the highest visits per 100 incidents.

  • The Bronx has the lowest.

  • This suggests that high-need boroughs may not be receiving equally accessible support.

Scatterplot: Incidents vs Client Visits

Figure 27

Key Interpretation

  • The pattern suggests a weak positive relationship.

  • Still, boroughs vary noticeably around the trend line.

  • Service engagement does not rise proportionally with domestic violence burden.

Discussion

Domestic violence resources are not evenly aligned with reported need across NYC boroughs. The Bronx shows the highest incident burden but the lowest service engagement relative to need.Staten Island shows much higher service engagement per reported incident. These disparities may reflect:

  • access barriers

  • transportation limitations

  • language barriers

  • fear of retaliation

Conclusion

  • The findings raise concerns about whether domestic violence resources are adequately meeting survivor needs across NYC.

  • In the highest-need boroughs, especially the Bronx, service engagement appears disproportionately low.

  • This is not just a statistical gap, but moreover it reflects real consequences for survivor safety, well-being, and long-term stability.

  • Improving access, visibility, and distribution of services is a public responsibility.

Final Takeaway

  • Higher reported need does not always correspond to stronger service engagement.

  • To better support survivors, NYC should evaluate how domestic violence services are distributed, promoted, accessed and resourced across boroughs.

Connect with the Students

Explore their work and connect:

πŸ”— NYC Open Data Student Gallery

πŸ”— NYC Open Data Student Gallery Github