10 Steps to Becoming a Data Scientist
August 1, 2016 3:15 PM
The information age has made untold amounts of data and information more easily created and available, but that means a melding of analytics and computer science has become essential in understanding and explaining it all. Thats where modern day data scientists earn their chops. Scott Zrebiec, Ph.D and senior data scientist at Valen Analytics, offers his thoughts on how to best break into the data science space. These tips will help understand the scope of the career, but is by no means an exhaustive list. As Zrebiec explains, data science is a profession that is constantly changing and reinventing itself.
Data science is not a stand-along topic
There can never be an academic course focused on data science as a standalone subject. Instead, it is a collaboration of finely-tuned quantitative and creative skills. There are obvious skill sets data scientists need in order to be proficient, including advanced mathematics (numerical analysis) and computer programming, but also some less obvious soft skills. This includes an aggressive need to pursue knowledge for knowledges sake, a healthy sense of curiosity, a self-starter attitude, and the ability to explain insights found from the data to others.
Good data science starts with good communication
Data tells a story, but it is the data scientists job to teach the language and communicate that story to others in the business that must base key decisions off of the insights. These skills are nearly as important as the technical ones, because they are what drive you to think like a data scientist, learn more about a field that is constantly changing, and show initiative to prospective employers. The skills can be gained through real-world experiences, but are usually ingrained in an individual. Something like curiosity is hard to teach, but is essential to success.
Lessons learned outside the classroom
While university courses are necessary for technical abilities, many of the skills my colleagues and I possess were attained beyond the classroom. Like many professions, internships are an excellent way to gain on-the-job experience, since they often teach you how to ask the right questions. At its most basic level, that is what a data scientist must always do, because without asking the right questions, you will find yourself shooting in the dark. It is also imperative to work in teams with both rookies and veterans, long enough to guide and point you in the right direction. This is also an important way to network, which helps in the long run, like with any field you are attempting to break into.
Scour online courses and contests
Some out-of-the-box ways to gain necessary experience include taking the initiative to scour online courses that may offer additional insights and skill sets. Something like the online courses through MIT (which have been great sources of knowledge for me) or Kaggle.com contests are a good way to both test your skills against peers, and let companies and organizations see how your mind works. These are excellent ways to set you apart during the job hunt.
Use the job interview as a learning tool
The interview process for data scientists is unique and something I am heavily involved in at my current position. Interviewers are always looking for candidates who can tie data and analytics back to business value, and many of our questions are a test to see how you think, problem solve, and take on challenges. While every specific industry is different, expect general questions followed by thought-provoking and in-depth questions in fields like computer programming and statistical modeling, as well as related fields you may not be familiar with, just to see how you think about a problem you werent expecting.
Think in an open-ended fashion
A general question that Ill often ask someone who doesnt have a formal computer science background is: what is the computational complexity of a sort? This is ultimately unimportant, but its also a simple way of knowing how deep someones knowledge on programming is. A more open-ended question would be how would you predict an avalanche? This is a good question, because it forces people to think about both what data to obtain and how to produce a prediction. Together, these help gauge the depth of a candidates knowledge and the ability to think on their feet.
Tie everything to the business
There is a growing sentiment in the business world that when data scientists are first starting out, they fail to possess an increasingly important skill: communicating technical information to the C-suite that is clear, concise and relates back to business results. Though they may be proficient in the technical aspects of the job, they have not yet learned how to tie that back to business interests, which at the end of the day, is the goal of their research.
Solve complex problems in simple terms
Communication (or lack thereof) will separate the good data scientists from the great ones, so it is critical to see how others in the field have parlayed their technical knowledge into effective communication. Can a potential candidate solve complex problems, but also write and communicate that complex problem and describe the takeaways without losing executives in the details? Many in the C-suite may not have the same technical experience or vocabulary you have, so translating that language is essential for receiving organizational buy-in and continuing data projects.
Take pride in proving yourself wrong
While being a data scientist is incredibly fun (if you're the naturally inquisitive type), it can become frustrating, since the true purpose of a data scientist is to continuously prove your own work wrong time and time again. Data science is about finding new answers to existing problems on a regular basis, and being right is equivalent to plateauing. That means tearing up your work constantly and attempting to find holes and logical fallacies at every step, criticize it from every angle, as well as being open to feedback and even criticism from others.
Recognize the power of the team
There will be many times a data scientist will try a technique that might not work well repeatedly just to verify and re-verify the results. Patience and teamwork are keys to success, as well as utilizing colleagues that might more easily spot something you may have missed. A group of data scientists can be smarter than one individual data scientist.