The US Special Operations community is a diverse group of highly trained men and women dedicated to the cause of "protecting and advancing U.S. policies and objectives across the international arena". The community is composed of Special Operations Forces (SOF) "tribes" from all branches of the Department of Defense to include the Navy SEALs, Army Green Berets, Air Force Pararesucmen, Marine Raiders, and jet and helo pilots, among other specialized units. This highly talented subset of the overall US population, at some point in their military career, makes the decision to "put down their ruck sack", "hang up their uniform" and transition into civlian life.
As a former SEAL turned Data Scientist I partnered with Elite Meet - a 501(c)(3) non-profit Veteran Service Organization dedicated to providing a post-military SOF network and connecting its members to leaders in the Corporate sector - to conduct an analysis on this specialized group of veterans to paint a holistic picture of who they are and where they are headed. This analysis addresses the following questions:
- Where does this post-military network live?
- What is the composition (by tribe) of the overall community?
- What are the descriptive charateristics of this group: age, education, years of experience,...
- How prepared is this group for the Corporate workforce?
- How has Elite Meet grown over time and through what channels?
I was granted access to an anonymized section of the Elite Meet (EM) membership database which consisted of 750+ entries of SOF veteran members. This initial dataset was curated from the EM Salesforce instance, which in turn was collected from individual membership intake forms at time of EM application. Because many of the intake form fields were free form, the dataset required an extensive amount of cleaning in order to conduct an accurate analysis. As an example, something as simple as inputting a "Current Location" oftentimes included several variations, as shown in the figure below:
To tackle this challenging (messy) dataset I used the following tools for the task:
- Data Cleaning: Excel + pandas combo
- Data Analysis + Plotting: Jupyter Notebooks, pandas, matplotlib
- Interactive Viz: Plotly
- Storage and presentation: Github + markdown
The dataset originally contained more than 50 features/columns, but after reviewing the data I was able to reduce the number of features to 32. It took 20+ hours just to clean this dataset before it was ready for a single line of analysis. One of the recommendations that I made to the Elite Meet organization is to change their intake form to include more drop-down menus and autogenerated responses, in order to eliminate the need to extensively clean the data for future analysis.
One of the first things that the CEO of Elite Meet wanted to know is, how are the members geographically distributed? One of the original ideas behind the Elite Meet brand was to establish regional chapters that would serve to support members within their respective regions. The CEO wanted to get a good sense of where to make a marketing push for regional chapters.
I decided that a heat map, which displays the denisty of points on a map represented by color intensity (heat), would be a good way of visualizing the geographic distribution of membership. I originally decided to work with Folium, which is a Python library with a Javascript backend that provides mutiple dynamic, interactive overlays. However, after working with the library I realized that the weights function was not working which is a known issue with the library. I ended up working with Plotly, which provides a great API for mapping interactivity. Below is the final result, clicking on the map opens an interactive window:
Anyone familiar with the SOF community would not be surprised by this distribution. Members are clustered around all of the major military bases housing SOF personnel - 1st Special Forces Group in Seattle, Navy SEALs and Navy pilots in San Diego and Virgina Beach, 5th Special Forces Group in Kentucky, 3rd Special Forces Group in North Carolina as well as the Raiders out of Camp Lejune, and the list goes on. To get a better sense of the East Coast distribution, which is a little busier than the West Coast map, I provided the following zoomed-in image:
- For an interactive solution to visualizing the national distribution of EM members click on the link below:
National Distribution
Next on the list was providing the CEO with an understanding of the actual makeup of the Elite Meet community, specifically by SOF "Tribe". There are many different SOF variants and members were given free reign to describe their primary job duties at the time of application and therefore exact tribal classification was not readily apparent to the casual observer. A better understanding of the community makeup allows the CEO to ensure that outreach efforts are balanced across the Defense service branches and that she can communicate an accurate picture of the community to potential hiring partners/managers.
I provided two diagrams:
- First chart provides the SOF Tribe breakout by service (Navy, Army, Air Force, Marine Corps)
- Second chart provides a holistic overview of the entire community at once
For those unfamilar with the several SOF variants that make up US Special Forces, some of the acronyms in the below charts might not make any sense, so I've created a brief non-exhaustive glossary below. Also, for standardization across graphics the following color chart will is used in keeping with traditional military coloring of the service branches:
- Army = Green
- Navy = Blue
- Air Force = Sky Blue
- Marine Corps = Red
- Joint = Purple (mix of branches)
This first chart breaks out the community by tribe within their respective service branches. For example, 46.7% of all SOF Navy personnel are SEALs while 13.7% are SWCC. No real suprises here with regard to the community makeup.
This second chart provides both a side-by-side bar chart comparison of raw counts, as well as a "Tree Map" which shows proportionality in a different visual style. Two points to note here.
- Given the size of the Army Special Forces, one should expect about a 3:1 ratio of Army to Navy personnel in these charts. But, given that Elite Meet was originally founded by a Navy SEAL and SWCC member, much of the initial community was comprised of Navy personnel given the word-of-mouth nature of recruitement. As you'll see in one of the following charts, Army personnel do start to pick up their numbers in more recent months, but even still, they are vastly underrepresented in the Elite Meet community given their associated active duty population size.
- Second point to note is that about 20% (1 in 5 members) of the community are fighter and helo pilots, which is far more than I would have originally assumed. I'm not sure what the total numbers are for the pilot communities within the Department of Defense, so perhaps 20% is right where they should be, but either way their healthy membership numbers indicates that the word is out within the pilot community that Elite Meet is a good network to be a part of.
It's now time to turn to the individuals that make up this community. The CEO wanted to know some basic statistics on the following:
- Years of Experience
- Education Levels
- Target Industries
The following graphs/charts begin to paint a picture of what a veteran from this community looks like, with some interesting suprises along the way.
To begin, we start with a distribution of the years of experience found within the community.
So how educated (formally) are the veterans from the Elite Meet community? Surprisingly enough (to me) - very.
I also thought it might be interesting to see which graduate programs this group is attending. With such a high rate of graduate education, I was curious to see if these degrees were obtained through "degree mill" programs. The word cloud below shows the opposite:
As an aside, I did a quick comparison of Degree attainment (Undergrad and Grad) broken out by branch. The Marine Corps and Air Force lead the way in both categories, in both cases trouncing the Navy. I have a Navy background so I wasn't terribly pleased to see these results...
| % of Undergrad Degrees: By Branch | % of Grad Degrees: By Branch | ||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
So where do these highly educated and well-experienced group of individuals want to work? Listen up Corporate America, below are the answers to the test:
- Consulting: 17%
- Entreprenuership: 12.3%
- Tech: 11%
This information is important in two ways.
- The Elite Meet CEO now has an idea of which industries/companies to target outreach efforts to when setting up interviews for Elite Meet members. The big consulting firms (Deloitte, Accenture, BCG, etc.) should be at the top of her list. All of this data can also be used for targeted marketing campagins which highlight the general traits of this talent pool directed at companies looking to hire.
- Hiring managers and recruiters who have access to this talent pool now have an idea of what this group of veterans are interested and can act accordingly.
As the final part of this analysis, I chose to provide the Elite Meet CEO with an idea of how the organization has grown over time, what the makeup of that growth looks like (from a service branch perspective), and which of the "marketing" channels are most effective in bringing in new members. First a look at growth over time from the last two years:
- Navy: 1
- Air Force: 2
- Army: 3
- Marine Corps: 0.33.
Meaning that for every 1 Navy member, we should see 2 Air Force, 3 Army, and 1/3 of a Marine Corps member. These ratios clearly don't hold up in the graph above. The Air Force has very low membership rates, considering their overall Active Duty SOF population, and the Army, even though they have the most members in the organization, should have close to triple their current numbers. The easiest explanation for this disparity is that Elite Meet was found by a Navy SEAL and SWCC member largely by word of mouth. So the inter-service cross pollination did not start taking hold until recent months.
