When you graduate from the United States Coast Guard Academy, you receive a commission as an active duty officer in the U.S. Coast Guard. As a newly commissioned officer, the Coast Guard’s number one priority is building your experience. And thus, you typically go to different operational units. Any assignments as a newly commissioned officer that may not be considered operational typically result in an officer being placed into a specific community with a well-defined officer specialty, or career path. And these first assignments would be unique and few and far between.

Through 2024, being a data scientist is not really a viable career path in the U.S. Coast Guard. Anecdotally, every Coast Guard data scientist has multiple examples of data scientists they know who fail to advance in the organization. What is arguably more alarming is data scientists do not have the clear career progression opportunities outlined for other officer career paths. And the result is a limited number of positions with substantial preliminary entry requirements (typically a Master’s of Science in some form of analysis), dwindling numbers of qualified personnel to fill them, and promotion/review boards that do not understand the positions and work. Subsequently, these promotion and review boards then do not adjudicate favorably on a data scientist’s behalf (data scientists largely fail to promote to senior levels).

So, I graduated from the Coast Guard Academy with a Bachelor’s of Science in Operations Research and Computer Analysis. And I reported to U.S. Coast Guard Cutter VIGOROUS as an Engineer Officer in Training. From VIGOROUS, I was assigned to a position within Surface Forces Logistics Center’s Medium Endurance Cutter Product Line as a Port Engineer. It was while I was assigned as a Port Engineer that I decided to further my education as a data scientist and pursued a Master’s of Science in Data Analysis, online, in the evenings and on weekends.

Ultimately, the efforts I put into continuing my education while I was a Port Engineer were noticed by the Coast Guard. As I finished a course that would bring me to thirty total graduate credits towards my MS in DA curriculum, I was selected for a Coast Guard funded post graduate advanced education opportunity, where I would earn a Master’s of Science in Operations Research. Of course I finished the MS in DA. And as I transferred from the Port Engineer position, I attended school full-time in pursuit of a MS in OR. Upon graduation from the MS in OR, I reported to my first data scientist position within the Coast Guard; I reported to the Office of Requirements and Analysis.

U.S. Coast Guard Office of Requirements and Analysis (CG-771)

I graduated from my MS in OR curriculum after the Fall semester of 2019. In early February 2020, I reported to the Office of Requirements and Analysis as an Analysis Officer. And I was honestly very happy in CG-771. I had a great command, and they enabled me to aggressively seek out problems I could help solve with math. Upon reporting I was able to contribute to a Safeguarding National Sovereignty in the Arctic and Antarctic study through mathematically determining and interpreting a fleet mix of heavy and medium icebreakers. And shortly afterwards, I was able to contribute to the Western Commerce Cutter acquisition by performing an optimization for the historical work completed by the inland construction tender (WLIC) class of cutters.

As I stated, I was very happy in CG-771. The COVID-19 pandemic hit, and still I was enabled to push through math and inform Coast Guard decisions. I had a command that was both trusting and supportive. And I was continually pursuing the passion that had driven me through two MS degrees to a job as an analyst. However, I was encountering a recurring problem.

Coast Guard Data

With any analysis, a data scientist will tell you the bulk of work is associated with data. Whether it is mining, cleaning, munging… an easy majority of an analysis is data. The data problems I encountered in the Coast Guard were living up to these sentiments. But there was something much worse going on.

There would never be an assumption Coast Guard data was ready to go. Every analysis idea starts the same way:

“Where is the data?”

“You don’t have data. Where can we get data?”

“Meh, we have data… but we have very low confidence in the data.”

With the last statement, an even more frustrating scenario would occur where after performing an analysis, the decision makers who were the primary stakeholders in the analysis may attack the analysis itself via reliability of the data; because they did not like the results of the analysis.

To an extent, everything I just highlighted is common to data scientists across the industry. Data is messy. An analysis does not provide the desired answer and is ignored. These situations are probably not very shocking. But I ran into something much more startling.

What happened is, I ran into decision makers who were open to (even enthusiastic for) analytic solutions. But because of what these decision makers were told by data scientists, they thought they could not support analysis (i.e. they could not provide data to support robust analysis). I remember listening in awe as a senior leader described how much he loved operations research graduates, how he had worked alongside some of his peers who were OR graduates and was completely amazed. And then my jaw hit the floor when he described how he brought different analysis ideas to different data scientists and was basically told they could not help him, because of the data.

The Coast Guard’s data was messy for sure. But there was a larger cultural sickness because of the messy data. Data Scientists were empowered to turn away customers on data quality alone. And similarly, customers were discouraged from an informed solution because they felt shame over a problem they had little to no control over resolving.

These views are mine and should not be construed as the views of the U.S. Coast Guard.