Data Science Degree in a Data Analytics World

The whirlwind of graduating and starting my real life as an adult post-school has been rather exciting and I think I'm finally settling in. For those who haven't been keeping up, after graduating UCSD last year (wow!) I have since moved to Las Vegas to start as a data analyst at Caesars Entertainment. To get the non-technical commentary out of the way quickly, Las Vegas has been such a fun city that is exceptional at two things: its proximity to nature and its world-class clubbing scene. So naturally, I've been finding myself doing both. We tried doing a single day going to Zion in the morning and the clubs in the evening, just to give you a sense of what's possible.

Data Science and Data Analytics

One of the big questions that I went into this role uncertain about was what exactly data analytics meant. I didn't want to discount the entire career path of data analytics just off the basis that the word "Science" ≠ "Analytics", and I will say the two things are somewhat further apart than I had previously thought.

I loved HDSI's data science major. It gave me exactly what I wanted: the intersection of math and decisions. Learning about decision stumps and semaphores and D3 may all have caused a little more stress at the time, but I could always separate the impending deadline doom from the appreciation of the academics I got to acquire. But how did this wonderful data science degree prepare me for data analytics?

The relationship between the knowledge I need to be an analyst and what I gained as a data scientist initially seemed somewhat loose. The analytics world is so rooted in the pipeline of SQL into Excel or Tableau, the old high school math adage of "when would I ever use calculus back propagation in the real world?" seemed woefully relevant. That being said, my brain had been folded into the shape of neatly organized rows and columns using pandas so I wouldn't say that I was necessarily under-prepared. The biggest hurdle for my brain was (and is) downloading all of the business context surrounding the data. Something that wasn't covered to a great degree in my degree was the careful connection between how the data is collected and how the data presents itself in the tables. Side note, this data is messy. I knew data would be messy in the real-world, but academia really is a safe bubble. Even the messy data is easily cleaned. At work, what is the relationship between the suffix _notadj and _notadj_old? You better use the right one!

Data Science in Data Analytics

After nine months of being at Caesars, with a little bit of luck and good timing, I have been promoted from an Analyst I to an Analyst II (that's 100% more analyst!). To allow myself a single clause to brag, nine months is much quicker than the typical 13-16 month roadmap that the rest of my peers preceding me followed. I think that despite its initial loose ties, I can cite my data science degree as a great reason how I was able to find myself here so quickly. Hopefully there will be more to come as well.

The high level concepts that analytics and data science bathe in are one in the same. Look at data and help us make decisions. With the help of many lovely managers, I have been given ample opportunity to demonstrate the value of the data science skills I possess. Causal and statistical analysis approaches that may have been just out of reach, I have managed to bring closer into the spotlight. Models and forecasts that highlight items that are the most important to evaluate allow the decision makers to do their same job, but in a more efficient manner. My ability to conceive of these approaches on my own, confidence in advocating for a potential solution, and speed in developing proof-of-concepts that excite the stakeholders, are all consequences of the things that UCSD's data science program has taught me directly.

There are several core concepts that I think have been most important in making me stand out (and one that wasn't taught at UCSD).

1. Being comfortable with tabular data—I don't have much to say, this point is so important that it's almost trivially obvious.
2. Strong model math understanding—The deep understanding of each model, how a given GLM may be useful, when to use a random forest or to upgrade to a boosted model, why DNNs may not be practical despite them being all the rage. It is the understanding of the foundation of each of these approaches that gives me confidence in using and advocating for them in rooms where people may not know much about the depths of models.
3. Data visualization and encoding information—A good product is one that looks good and feels good to use. I have found that knowing how to emphasize the point of a dashboard or a report sells the purpose better than just a model output on its own. One project was all about the model forecast. To motivate this, I added a dotted forecast line as well as confidence intervals in a way that aligns the stakeholders' thinking with the thinking of the model.

The final piece I want to emphasize I feel deserves its own section entirely. Is it more core than the data science technical skills? Probably not, but it is the core lubricant that allows the gears of my degree to turn smoothly.

Communication is Key

I am lucky that my own interests and personality had led me to interpersonal jobs throughout college. Interpreting at aquariums, leading orientation groups, tutoring data science, all jobs that forced me to learn how to communicate well and effectively to groups of all different types.

The ability to present and communicate what you've done, what you're doing, and what you want to do have proven to be vital in getting the most interesting projects and then really making them shine. Knowing what parts of each project are interesting to any audience has been something that made me stand out within my first nine months. Presentations to the analytics team—what can I take away and use? Presentations to stakeholders—how does this solve my problem? Presentations to transfer a product—what are the highlights, pitfalls, and areas for improvement? I could go on and on.

Being good at presenting has given me a lot of goodwill. Most people don't actually know what I've been up to within the department, but they know I've given good presentations. Murmurs of having great team-wide presentations make me feel personally redeemed as well as have given me that one step up among the team. If I had to recommend one mid-semester job, I would tell everyone to tutor for their data science or math class.

Parting Words

Nine months in industry isn't a lot. I am still a child in this career world. There is lots to learn and I hope to learn it. That being said, formal education can quickly be lost after not using those skills for a long time. I had developed some nice skills! I hope not to forget them and to be able to remember what is important to continue fostering from college into the working world. My degree may not have set me up to do the exact role of "data analyst", but it has opened far more doors than if it did.