Socio-Economic Effects on COVID-19 Mortality Rates

A Machine Learning Approach

📋 Abstract

This research investigates the relationship between various socio-economic factors and state-level COVID-19 mortality rates across the United States. Using machine learning regression techniques, the study analyzes data collected from 2020-2023 to identify significant correlations between social and economic indicators and population loss due to COVID-19.

🎯 Research Objectives

The primary goal of this study is to determine which socio-economic factors most significantly influenced COVID-19 mortality rates at the state level, providing insights that could inform future public health policy and pandemic preparedness.

📊 Dataset

Data was collected from reliable government and institutional sources spanning three years (2020-2023):

Variable	Source	Link
Poverty Rates	USDA Economic Research Service	View Data
State Population	USDA Economic Research Service	View Data
Average State Income	Federal Reserve Economic Data	View Data
State Health Rankings	Forbes Advisor	View Data
COVID-19 Deaths	Statista / CDC	View Data

Additional Data Files Included:

• Provisional_COVID-19_Deaths_by_Place_of_Death_and_State_20240124.csv - CDC provisional death data
• State_Populations.csv - State-level population data
• Percentage-in-Poverty.csv - State poverty percentages
• average-income-by-state-2024.csv - Income by state
• united-states-by-density-2024.csv - Population density metrics
• h08.xls - Health insurance coverage data

🔬 Methodology

This study employs multiple machine learning regression techniques to analyze the relationship between predictor variables and COVID-19 mortality rates:

• Least-Squares Regression - Baseline linear regression model
• Multiple Least-Squares Regression - Multi-variable regression analysis
• Ridge Regression - L2 regularization to prevent overfitting
• Lasso Regression - L1 regularization for feature selection

📈 Key Findings

Findings can be found in the research paper attached to this file.

🏫 Acknowledgments

This research was conducted as part of the Belmont University SURFS and presented at the 2024 Summer Undergraduate Research Fellowship Symposium (SURFS).

---

Project Status: ✅ Completed (Summer 2024)