This repository contains the full analysis, cleaned data, statistical modeling, and written report for the project “Impact Analysis of Age and Gender on COVID-19 Mortality Rates.” The study examines how age and gender shape COVID-19 mortality patterns in the United States using CDC public health records. The analysis integrates statistical inference, data visualization, and baseline machine learning models to evaluate demographic risk factors.
The complete written report is included in: annotated-final.pdf
Understanding how demographic factors influence COVID-19 mortality is essential for designing targeted public health interventions.
This project investigates three core questions:
- How strongly age is associated with COVID-19 mortality
- Whether gender differences are statistically significant
- Whether simple predictive models can capture mortality patterns within the highest-risk age group
The project includes rigorous data cleaning, exploratory data analysis, hypothesis testing, and baseline machine learning.
COVID-19 mortality increases sharply with age. The age groups 65–74, 75–84, and 85+ show the highest mortality, with 85+ being the most vulnerable.
Although males exhibit higher observed mortality,
statistical testing shows no significant gender differences within age groups (p > 0.05).
Models tested:
- Logistic Regression
- Decision Tree
- Random Forest
Accuracy was low due to high variability and limited features, indicating that more complex predictors are required.
- Selected key variables
- Standardized categorical encodings
- Removed missing values
- Produced cleaned dataset (
cleaned_covid_mortality_data.csv)
- Mortality by age
- Mortality by gender
- Boxplots and bar charts
- Identification of high-risk groups
- Two-sample t-tests within each age group
- 95% confidence intervals
- Effect-size evaluation
- Training on the 85+ subgroup
- Evaluation using accuracy, confusion matrix, and classification report
- Age is the strongest determinant of COVID-19 mortality
- Gender differences are not statistically significant
- Predictive models fail to capture mortality complexity with demographic features alone
United States Centers for Disease Control and Prevention (CDC)
Provisional COVID-19 Deaths by Sex and Age Dataset
Shuozishan (Doris) Wang
B.S. Applied Statistics, University of Washington