Flu Death Prevention

I was initially drawn to this project out of a curiosity about the intricacies of the U.S. medical system, particularly in terms of population correlations and medical staffing rates across states.

The primary goal was to uncover predictive factors for flu-related fatalities, facilitating more efficient medical staff allocation nationwide. I scrutinized variables such as population, geography, and vaccination rates to identify high-risk groups. You can explore the comprehensive dashboard and presentation, which accompany my methodology bellow.

U.S. At Risk Population for Flu Death Map

This project took about 30 to complete and provided valuable insights into how different states approach flu risk management and early prevention. It underscores my proficiency in data analysis to convey essential insights visually through Tableau's powerful tools, enabling data-driven decision-making.

The Steps I Took:

  1. Translated business requirements into a project plan

  2. Cleaned Data for clarity and consistency using Excel

  3. Merged data sets into one

  4. Derived new variables

  5. Explored data using Tableau

  6. Presented custom Tableau dashboard via video recording.

Tools I Used:

  • Tableau

  • Excel

Why I took these step:

  1. Planned the analysis to meet the needs of the stakeholders

  2. Ensured analysis was correct with error free data

  3. Simplified data sets into one main data frame for ease of use

  4. Added depth to analysis by extrapolating the data

  5. Found useful insights for primary and secondary objectives

  6. Shared insights with stakeholders in a clear manner

Challenges:

  1. Before merging the data sets I did upload the data frames to Tableau thinking I would just merge them there using the columns 'state' as the primary key. I quickly realized though that this was making things more complicated than it needed to be especially when I was working with horizontal and vertically organized data.

    So I went back and reformatted the data using the pivot tables and transpose.

    After I did this I no longer needed to set primary keys in Tableau which helped to make the process easier. In the future I expect to only use Tableau's key setting feature with data already in a pipeline.

  2. This one is more of a caveat than a challenge but Tableau's predictive feature is really quite one dimensional. I've always thought you should be able to load 2-3 variables and pick your own weight for the resulting prediction. In the future this is something I'd like to explore with machine learning perhaps.

The dashboard is not available on mobile view. Please watch the video instead.

Full Presentation

Data Sources:

  1. Flu Deaths

  2. Population by Geography

  3. Flu Test Counts

  4. Flu Shot Rates in Children

Retrospective:

This project used my excel skills heavily to effectively clean and format the data into a usable manner. Once, I had wrangled the data though it was a lot of fun to visualize it in Tableau and create a thought provoking analysis with it. Using guiding questions as the headers created a thought provoking progression of the analysis instead of a fixed, answer-first, state of mind which tends to kill the thinking process.

Going forward, I want to use predictive forecasting models with Sikit machine learning to create more realistic forecasts. This would be a separate multi variable project that would let me better understand the foundations of forecasting.

Overall, I am very happy with how this project turned out both analytically and visually. I provided depth and breadth to the objectives and sparked an interest in forecasting that I otherwise would not have. This analysis was also eye opening in how states relate to one another in the U.S for Flu deaths and prevention. It was a worthwhile analysis.