DSAN 5100 - Final

Project Overview

This project investigates the relationship between public health metrics and housing risk scores in various districts. By analyzing vaccination rates, death rates, uninsured percentages, and housing-related factors, the study seeks to identify disparities and potential areas for policy intervention.

---

Research Question

What are the key factors contributing to high housing risk scores across districts, and how do public health and socioeconomic indicators correlate with these risks?

---

Datasets

The dataset used in this project includes housing risk scores and a variety of public health and socioeconomic metrics:

• Source: Dataset
• Features:
  • COVID-19 Vaccination & Death Rates
  • Percent Uninsured
  • Crowding Percent
  • Area Median Income Percent
  • Percent People of Color (POC)
  • Rent Burden Percent
  • Eviction Filings Rate
  • Tenant Cases Rate
  • Housing Violations Rate
  • Unplanned Outages Rate
  • Change in Median Gross Rent
  • Change in Sale Price
  • Number of New Unaffordable Units
  • Rate for Foreclosure Filings
  • Share of Nonbank Small Home Loans
  • Total Housing Risk Score

---

Key Findings

1. Public Health Metrics and Risk Scores

1. Vaccination Rate vs. Risk Score:

   • Trend: Negative correlation.
   • Insight: Higher vaccination rates correlate with lower housing risk scores, indicating better public health infrastructure.
   • 

2. Death Rate vs. Risk Score:

   • Trend: Positive correlation.
   • Insight: Districts with higher death rates tend to have higher risk scores, reflecting potential disparities in healthcare access.
   • 

3. Uninsured Percent vs. Risk Score:

   • Trend: Positive correlation.
   • Insight: Higher uninsured percentages are linked to increased housing risks, suggesting the importance of health insurance coverage.
   • 

2. Borough-Level Comparisons

• The Bronx (BX) has the highest average risk scores and death rates.
• Manhattan (MN) leads in vaccination rates and has the lowest death rates.
• Queens (QN) has the highest uninsured percent.

Visuals:

• Risk Scores by Borough:
• Death Rates by Borough:
• Vaccination Rates by Borough:
• Uninsured Percent by Borough:

---

Statistical Methods

1. Hypothesis Testing

1. T-Test for Vaccination Rates:

   • Districts with above-median vaccination rates have significantly lower risk scores compared to those below median rates.
   • p-value: 0.00017 (statistically significant).

2. ANOVA for Uninsured Percent:

   • Significant differences in risk scores exist among groups with low, medium, and high uninsured percentages.
   • p-value: 1.39e−05 (highly significant).

2. Correlation Analysis

• Strong correlations exist between public health metrics like limited English proficiency and rent burden.
• Figure:

---

Data Visualizations

Public Health Metrics

• Vaccination Rates:
  

• Death Rates:
  

• Uninsured Percent:
  

Risk Scores

• Top 20 Districts by Risk Scores:
  

---

Presentation Structure

1. Introduction:
   • Explain the research question and its importance.
2. Exploratory Data Analysis (EDA):
   • Share general findings and insights.
   • Time Allocation: 3 minutes.
3. Public Health Analysis:
   • Detailed analysis of vaccination rates, death rates, and uninsured percentages.
   • Time Allocation: 3 minutes.
4. Statistical Results:
   • Explain hypothesis tests, correlations, and implications.
5. Conclusion:
   • Summarize findings and policy recommendations.
   • Highlight actionable takeaways.

---

Next Steps

1. Expand Dataset:
   • Include individual complaint data to analyze specific housing issues.
2. Advanced Analysis:
   • Implement Bayesian methods to estimate probabilities of high-risk outcomes under various conditions.
3. Policy Recommendations:
   • Focus on reducing uninsured rates and improving vaccination coverage in high-risk districts.

---

Authors and Contacts

For any inquiries or further collaboration, please reach out to the project authors:

• Alivia Castor
• Amanda Hao
• Satomi Ito
• Mandy Sun
• Jeffrey Kolbeins

---