Tableau Project

Understanding Flu Season

My Role
Data Analyst
Timeline
October 2022 - December 2022

Understanding Flu Season

For the entire analysis, the Interim Report is available here

“Preparing for Flu season” dashboard is available on Tableau Public here

Overview

A medical staffing agency covers all hospitals in each of the 50 states of the United States, and the project will plan for the upcoming influenza season. This agency provides temporary workers to clinics and hospitals on an as-needed basis.

Goal

Determine when and how many staff to send to each state.

Tools

  • Excel (full analysis)
  • Tableau Public (final presentation)

Data

  • Influenza deaths by geography, time, age, and gender. Source: CDC
  • Population data by geography. Source: US Census Bureau

Data Limitations

  • Firstly, we don't have data from the past five years (2017-2022) in both of the data sets.
  • Even if it is possible to predict prospective flu seasons using past influenza data for planning purposes, this may still represent a bias.
  • We lack data: in the CDC data collection, all statistics indicating zero to nine (0-9) deaths or births are censored for privacy reasons, whilst in the USA census data set, 13% of the total data points are missing.

Skills applied

  • Data Profiling & Integrity
  • Data Transformation & Integration
  • Statistical Hypothesis Testing
  • Forecasting
  • Storytelling with Data presentation

Business questions

  • Does the flu season affect some areas more than others?
  • How long does the influenza season typically last? When does it begin and end?
  • Which demographic groups are thought to be more vulnerable?
  • Which states have a higher concentration of vulnerable people?

1. Statistical Hypothesis Testing

Research Hypothesis: If the patient is older than 65 years old, the chance of death after contracting the illness is higher.

I looked at the connection between the population over the age of 65 and the number of influenza deaths in that age group to test my hypothesis. This descriptive analysis shows a very strong coefficient (0.94) that suggests older patients are more likely than younger ones to die from the illness.

I performed a t-test with a high significance level of 0.05. Since I assumed directionality—that people over 65 have a higher risk of dying from influenza than younger ones — I conducted a one-tailed test. I actually got a p-value that is lower than the alpha (4.23438102120569E-124> 0.05). As a result, I was able to reject my null hypothesis (people 65 years of age and older have the same or a lower probability of dying from influenza than people 64 years of age and younger) and conclude with 95% confidence that the 65+ population has higher influenza fatality rates than those 65 and younger.

2. Spatial Analysis

Which states are most affected by the disease?

As we can see from the visualizations, the five states most affected by influenza deaths are:

  • California
  • Florida
  • New York
  • Pennsylvania
  • Texas

The states that may need extra staff are California, New York, Texas, Pennsylvania, and Florida. In particular:

  • California has the largest older population and the highest death rate.
  • New York has the highest death rate.
  • Florida has the second largest older population.

3. Temporal Visualizations & Forecasting

The visualization below shows that the death rate rises every year starting in November and peaks in January.

Based on historical data, we can also forecast for the next year (2018) as follows:

  • The number of influenza deaths may slightly increase in New York and Texas while decreasing in California, Florida and Pennsylvania.

4. Conclusions

The vulnerable population, especially those over 65 years old, is more likely to die after contracting the disease. For this reason, they require greater attention and consideration.

When is influenza season?

The flu season normally runs from October to March, peaking in January. The climate frequently has an impact on this and can differ from state to state. That is also a crucial factor when selecting where to dispatch extra staff.

Recommendations

In light of criteria like the vulnerable population and the level of need, the majority of extra staff should be sent during the winter. In particular, states such as California, Florida (which have the largest and second-highest elderly population, respectively) and Texas may require more flu shots and appropriate medical care.