The Correlation of Disease and Travel Frequency In The US

Alex Shen, Olivia Liu, Yahan Yang

September 2024

Carolina Data Challenge

Introduction

01

Datasets in Use

Disease and Travel

SELECT

Provisional Death Counts for Influenza, Pneumonia, and COVID-19: 2019-2023, Health Science

AND

Trips By Distance : U.S Department of Transportation: Bureau of Transportation Statistics Provided by Maryland Transportation Institute and Center for Advanced Transportation Technology Laboratory at the University of Maryland

​

Why Supplement?

• We wanted to see how different diseases affected the frequency of trips for US citizens with 4 years of data
• Local, Interstate, Long Distance

Results

02

Findings and Impacts

Data Visualization

We transformed our databases into graphs

relating deaths caused by the diseases and the amount of travel based on the trip distances.

​

• X-Axis: Deaths caused by Covid 19 or Influenza
• Y-Axis: Number of Local, Interstate, or Long Distance trips

CA Disease Deaths vs Local Travel

Scatter Plots with the Line of Best Fit

NC Disease Deaths vs Interstate Travel

Scatter Plots with the Line of Best Fit

WY Disease Deaths vs Long Distance Travel

There was no known influenza deaths in Wyoming from 2019 to 2023

Conclusion

—------------------------------------------

—Disease Correlations—

The number of deaths caused by Covid 19 has a negative correlation with most types of travel frequencies of people in certain U.S. States while the number of deaths by Influenza does not.

—Travel By Region—

Major states in the U.S. such as California, New York, and Florida were more heavily impacted by disease, while rural areas such as Wyoming saw decreased trip frequencies on a much smaller scale.

—Travel by Distance—

The Disease vs Local Trips plots showed higher magnitudes of negative slope compared to other distance of travel.

Impact

Business and Social Anticipation (Local/State)

• Recognize the rise of disease deaths ahead of time to implement policies for healh officials and other sectors
• Local transportation should anticipate decreased passenger counts during widespread disease, and could instead service as mobile health units, bringing testing or vaccination services to communities that are underserved or unable to travel to healthcare facilities.
• Businesses can pivot to online platforms to maintain operations if there are large counts of disease by death. They can implement delivery services or turn to subscription models to ensure consistent revenue.
• As humans are social animals, traveling less means less community and social engagement, which can lead to isolation and distress. Expanding access to mental health hotlines and crisis support services ensures that individuals in distress have someone to reach out to. Encouraging people to stay physically active and promoting mental health education are options too.

Business and Social Anticipation (Long Distance)

• While this analysis does not incorporate direct economic or policy data, the observed travel patterns during the COVID-19 pandemic deaths compared to influenza deaths suggest a more profound societal response to COVID-19, likely influenced by a combination of factors beyond the scope of this study. These factors may include government restrictions, social distancing measures, and public health policies.

​

• Covid 19: It is predicted by our model that people in certain States of the U.S. might be less inclined to travel if deaths rates of Covid increases, thus, airlines can potentially schedule less flights in reflection to conserve resource uses.

​

• Influenza: It is indicated that is no major change in people’s travel plans during influenza outbreak seasons as infected people increases, therefore, airlines can potentially plan to prepare more sanitation measures to ensure the safety of passengers and their employees while keeping the spread of the disease under control.

​

Process

03

Parameters and Manipulations

Start Week: Split into Year and Month

​

• Selected States
• Did not use Pneumonia because it is too closely related to Covid 19

Treating Null Values

Average_Death_Counts by States

• Sort by Year, Month, and State.

Parameters and Conversion Continue

Trips_Per_Month_By_States

• The original data contains up to 600,000,000 values.
• Local, Interstate, and Long Distance determined using miles per trip given in the dataset.
• Sort by Year, Month, and State.

​

​

Tool and Techniques

Things we used to make meaning out of data

SQL and Superintendent.app

• SQL for data wrangling and cleaning.
• Superintendent as the Integrated Development Environment (IDE) for SQL.

R, R Studio, and Data Visualization

• Created a Dataframe for each state and their respective values: type of disease and trip distances.
• Performed Linear Regression on each case and graphed the Normality Plots: Was not normal, decided to use boxcox.
• Boxcox better normalized the data and helped transform our database.
• Removed any Outliers with a High Cook’s Distance.
• Fitted the linear regression line again, and plotted each state’s values in Scatter Plot Graphs.
• Calculated the Confidence Interval for the slope.

Workflow

Abnormal Normality

Confidence Interval

Challenges

The obstacles we faced along the way

Thank You!

Q&A

CREDITS: This presentation template was created by Slidesgo, and includes icons by Flaticon, and infographics & images by Freepik