Home » Featured, Who Inspires You

Q&A: What Skills Do You Need to Succeed at Your Job?

1 September 2018 1,819 views No Comment
What skills do you need to succeed in your job as a statistician? We asked some of our members this and other questions and here is how they responded.

Bob Komara
Data Scientist at MSX International

 

 

 

What or who inspired you to be a statistician/data scientist?

I was well on my way to a PhD in physics with undergraduate degrees in math and physics when life got in the way. I realized I did not want to be an academic and many others were leaving academia for high-paying industry jobs. So, after completing a master’s and earning the infamous ABD title (all but dissertation), I moved to a Colorado ski town and started a new career managing high-end construction projects. I thought I would be able to bring my scientific background to bear on commercial problems and, to some extent, this was successful. But it wasn’t enough to keep me engaged. I kept reading new research and was very interested in the beginnings of the replication crisis and poor scientific reporting/communication being brought to light by different skeptic communities, and this was a distraction.

After a second major career detour that involved working at an aquarium store (I had one of the most beautiful coral reefs in Colorado in my bedroom); being a fly-fishing guide in and around Rocky Mountain Park, Estes Colorado; and launching and managing a state chapter of a national charity aimed at connecting children and their parents back to nature and each other through fishing called Fishing’s Future, I met my future wife (an academic psychologist) and we started a budding romance that involved an inordinate amount of heated discussion about data analysis, statistics, linear models, etc. She inspired me to go back to school and get a master’s degree in statistics so I could turn my “armchair” data analysis critiques into something more productive and, hopefully, lucrative. It didn’t hurt that she was awarded an important postdoc in Switzerland and I needed an official reason for residency to be able to live abroad with her. An incredible bonus was that while most things in Switzerland are expensive, graduate school tuition only cost me about 865 CHF per semester, which was an incredible bargain at about $930.

How did you end up in your current position?

After three beautiful years living in Switzerland, a master’s degree, and a marriage, we moved to the UK where I participated in a Knowledge Transfer Partnership (KTP) program for my first job as a data scientist after graduation. A KTP is a joint effort between the UK government, a university, and a commercial business that aims to link academic and business progress to spur innovation. I successfully applied my master’s education in spatial statistics/Gaussian Process regressions and R training with the help of academics at the University of Essex at Objective IT (a database and software development company). Together, we built a successful data science capability in the company by delivering successful products like including classifiers for highly imbalanced data sets and product recommenders for Europe-wide clients. After the successful and award-winning (Best KTP Associate University Essex 2017) two-year project completed, I accepted a new position at MSX International to apply machine learning techniques to complex and interesting problems in the automotive warranty industry.

Name a few specific skills you need to do your job.

  • Statistics
  • Communication
  • SQL
  • R
  • Vision to see the end goal

Data scientists must provide value. Many people rightfully include it as one of the big data “Vs” now.

What skill would you like to learn to be better at your job?

I am a believer that if you aren’t analyzing images, text, or video, then the business value of training interpretable machine learning models with good feature engineering exceeds the benefits of deep learning (for now), but want to take advantage of the power of LSTMs for time series forecasting.

What is the most exciting part of your job?

Getting good test results back on a deployed model and having everyone on the team celebrate the shared success!

What career advice would you give your younger self?

I wish I would have spent more time exploring options and taking risks earlier. I kept pursuing physics because I was good at it, but not necessarily because it was my passion. A better understanding of the exploration/exploitation dilemma would have given me better focus.

Name one or two favorite blogs or books you have read and would recommend to others.

I love reading R-bloggers. It is an aggregator of several hundred R blogs, and I love digging into tutorials and learning about new R packages featured there.

Two formative books I must mention are Innumeracy: Mathematical Illiteracy and Its Consequences by John Allen Paulos and The Demon-Haunted World: Science as a Candle in the Dark by Carl Sagan. Both were massively influential in blending my mathematical abilities with critical thinking skills and bringing them to bear on the question of data analysis in particular and science communication in general.

Scarlett L. Bellamy
Professor and Director of Graduate Studies in Biostatistics, Drexel University Dornsife School of Public Health

 

 

 

What or who inspired you to be a statistician/data scientist?

My grandmother inspired me to become a doctor (MD) when she was diagnosed with breast cancer while I was very young. I promised I would learn how to ‘fix’ her, but she passed away much too soon for me to fully execute my plan. I had always loved math, so in my mind, when I discovered biostatistics, it seemed the perfect compromise! I could be an ‘applied mathematics’ doctor instead!

How did you end up in your current position?

After spending 15+ years at the University of Pennsylvania—my first position after graduate school—I was looking to take on new challenges in a new position just after the school of public health at Drexel had gotten a new dean (Ana Diez Roux) and chair of the department of epidemiology and biostatistics (Leslie McClure). Drexel was also planning to launch a new PhD program in biostatistics, which seemed like a great opportunity to pursue, so I did! Both Penn and Drexel have been great fits for me at the respective stages of my career. Wouldn’t change much about my career path.

Name a few specific skills you need to do your job.

Besides the obvious (strong statistical knowledge and knowledge of statistical computing software), clear and confident communication skills. And as you take on more administrative responsibilities, leadership skills are also essential. Keeping on top of the literature and [having a] willingness to learn new things is also critical.

What skill would you like to learn to be better at your job?

I would like to learn more about visualization and exploring interesting data science/statistics intersections. Also, as I am taking on more administrative responsibilities, learning strategies to make me an effective leader is also becoming increasingly important.

What is the most exciting part of your job?

Every day is a new challenge, which is exciting! I love the autonomy and freedom to pick and choose problems I find interesting to work on. I equally enjoy the many ways I get to interact with students at all stages. They have so much energy and lots of great ideas!

What career advice would you give your younger self?

Take risks; confidence work is constant; and there are many ways to be great and yours/mine may or may not look like anyone else’s!

Name one or two favorite blogs or books you have read and would recommend to others.

Most of my reading is strictly for fun! I have very strict work/no-work boundaries. I recently finished We Were Eight Years in Power: An American Tragedy by Ta-Nehisi Coates and am currently reading The Awkward Thoughts of W. Kamau Bell by W. Kamau Bell. I should also plug my chair’s blog, StatGirl, which I also follow!

 

Julie Novak
Senior Data Scientist, Netflix

 

 

 

What or who inspired you to be a statistician/data scientist?

Funny enough, an Economist article. I was an undergraduate student at McGill University, where I was studying theoretical and applied mathematics. I was nearing the end of my bachelor’s degree and thinking about next steps. All I knew at that time was I enjoyed mathematics and problem-solving. I have fond memories of brainstorming solutions to challenging problems with friends.

The summer before my senior year, I came across an article from The Economist. It predicted that, with technological advances, massive data will be collected across all companies. There will be great demand to analyze and make sense of this data. The opportunities will be endless and exciting. Statistics will lead to unimaginable discoveries and drive decision-making in medicine, marketing, finance, tech, etc. The article also mentioned the work would be highly collaborative, which was another appeal for me. I am very social, so a job where I would have to frequently work with others was ideal. This role would also give me an opportunity to continue learning about new areas throughout my career. I then applied to PhD programs in statistics and went to the University of Pennsylvania to pursue my graduate degree.

How did you end up in your current position?

By chance! Prior to joining Netflix, I was a research staff member at IBM Research and not looking for a new job. This was the first company I worked for after I finished my PhD, and I was fortunate to have a brilliant mentor and manager, Yasuo Amemiya, throughout the two and a half years I worked there.

My current director, Nirmal Govind (who I did not know at the time), sent a note to my personal Gmail, asking whether I would like to have a conversation about data science at Netflix. I was flattered and curious about Netflix, since it is known to have one of the strongest data science departments in the tech industry.

Nirmal told me about the problems his team works on. He also sent me some Netflix Tech Blog posts (which I highly encourage those interested in learning more about Netflix to read). These articles gave me a clear understanding of what the team does. One of their focus areas is improving the quality of the streaming video service to its 100+ million members in more than 190 countries. They do so by building machine learning and statistical models, as well as running A/B tests at massive scale.

As I was going through the interview process, I was consistently impressed with how technically strong and personable my interviewers were, so when I was given the chance to become a senior data scientist on the team, it was an opportunity I could not miss. I currently develop nonparametric statistical methodology to improve the way we analyze A/B test results.

Name a few specific skills you need to do your job.

Besides statistics, the key skills needed for my job are “learnability,” coding, and clear communication.

Statistics was the only skill from the list I expected to be important prior to working in industry. Even so, the methodology one learns in school may not be relevant for the work they will do. For example, my PhD thesis was in Bayesian statistics, whereas I focused on forecasting at IBM Research and experimentation at Netflix.

This ties in closely with the second skill on the list: learnability. Every time I switched roles or companies, there were required tools I had not used before. For example, when I started as a data scientist at Netflix, I had to upskill on Python, SQL, Tableau, and R Shiny. Being able to learn new concepts and tools on the job is essential.

As for coding, I need to build, test, and debug models efficiently and effectively. Some of the code I develop is then used by other data scientists and stakeholders in the company, so it cannot break in production. I often work on projects with others, so the code needs to be understandable by others and reproducible.

Finally, the ability to explain your work to other specialists and nonspecialists is crucial. For stakeholders and colleagues to use your methods and trust your results, they need to understand how you got there. By being a clear communicator, you develop stronger relationships and, in turn, your stakeholders come to you with more problems and rely on your data-driven solutions.

What skill would you like to learn to be better at your job?

I cannot think of skills I need to learn from scratch; however, there are certainly ones I want to improve: coding, presenting, and writing. I did not know how fundamental each would be for statisticians. I wish I had been developing them for a longer time! For example, if you build an incredibly powerful model but others cannot understand what you did, it may lose the impact it deserves.

What is the most exciting part of your job?

I would say there are two: impact and stunning colleagues. It’s gratifying to work in a highly data-driven company, where decisions are made based on statistical results. If I have a hypothesis, I can run a large-scale A/B test to see whether it improves the streaming service for our customers. If the results show an improvement in streaming quality, the company will move forward with the change. An equally exciting part of my job is working with brilliant people every day.

What career advice would you give your younger self?

I would worry less about getting started in my career, especially while I was a PhD student. There is plenty of time after graduation to develop as a data scientist. School is only the beginning. Over the course of my career, I will have different data science roles and work at different companies. I should be patient with myself and work slowly toward my longer-term goals. I would also spend more time developing the skills I mentioned above (coding, presenting, and writing).

Name one or two favorite blogs or books you have read and would recommend to others.

I recommend the book Everybody Lies by Seth Stephens-Davidowitz (an economist, former Google employee, and New York Times writer). The author uses internet data and statistical modeling to try to understand how people think. This book is also a great way to demonstrate to your friends and family how data and statistics can help us answer questions about humanity in ways we could never do before.

I also enjoy reading Andrew Gelman’s blog, Statistical Modeling, Causal Inference, and Social Science. Gelman is a professor in the department of statistics and department of political science at Columbia University, and I would recommend following his blog.

 

Lisa Lix
Director of the Data Science Unit in the George and Fay Yee Center for HealthCare Innovation

 

 

 
In honor of this year’s JSM being held in Canada, the ASA’s Health Policy Statistics Section (HPSS) went to Winnipeg in the great province of Manitoba to interview longtime HPSS member Lisa Lix. Lisa is professor in the department of community health sciences at the University of Manitoba. She is also director of the Data Science Unit in the George and Fay Yee Center for HealthCare Innovation.
 

The George and Fay Yee Center sounds like an exciting place!

Yes! It is a nifty place, designed to bring transformation to health care in the province. Its eight platforms focus on developing and implementing evidence-based initiatives that improve care and outcomes for all Manitobans. My unit employs about 30 people, including faculty, technical staff, and trainees from biostatistics, bioinformatics, clinical research, and computer science. We develop and analyze data sources around the idea that health care needs to be more patient oriented.

How does your work dovetail with the Canadian health system?

Well, the Canadian health system is not actually universal health care; each province and territory gets an envelope of funding, which is delivered at the provincial and territorial level. What is done is determined by the province. Ultimately, the single-payer health system looks quite different in the different provinces and territories. Unfortunately, this applies to data as well. Hospital records are standardized by the Canadian Institute of Health Information, but physician records and outpatient care databases look different in different provinces. So, there is a strong push for cross-provincial and cross-territorial studies by the Canadian Institutes of Health Research.

It seems like there are interesting funding models based on the partnerships between the provincial governments and health researchers.

This is true. We receive funding from the Canadian Institutes of Health Research’s SPOR (Strategy for Patient-Oriented Research). In fact, the Data Science Unit grew out of SPOR funding. SPOR wanted to fund methods hubs in each province, including Manitoba. There is a strong connection between provincial/regional health care decision-makers and research, facilitated by funding models like SPOR.

What’s it like living in Winnipeg?

Winnipeg is a fantastic city to live in, very multicultural. The weather may be cold, but the people and culture are not. Twenty percent of the population is indigenous (First Nation and Métis). So, there is much to do in terms of activities that have a multicultural focus. Come visit!

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)
Loading...

Comments are closed.