'We Are Likely 3-5 Years Out From Advanced Analytics Being Critical To The Viability Of A Company'

Interview with Walter Storm, Chief Data Scientist at Lockheed Martin


Walter Storm is the chief data scientist for Lockheed Martin, guiding the corporation toward the application and extraction of value from emerging techniques in data analytics. Walter also co-leads the Big Data Analytics community of practice; is the co-chair for the decision support fellow’s action team; created the Data Analytics Pipeline Program; and provides consultation on advanced analytics and data science investments and activities across the corporation. Walter has also worked for Ford Motor Company, Proctor and Gamble, Tobyhanna Army Depot, and BBT Capital Management LLC.

We sat down with him ahead of his presentation at the Business Analytics Innovation Summit, taking place in Las Vegas this January 25 & 26.

What first sparked your interest in data science?

My data science journey started ca. 2001 with an exploration of probabilistic experiments aimed at certifying safety-critical software for flight worthiness. It started with the status-quo assertion that you cannot prove a 'shall not' statement through testing - it started before 'data science' was even a thing. As an aerospace engineer, I was trained to solve for the unknowns - to find the answer. I was trained to design the experiments, instrument the models, collect the data, and solve the problem. As software became more complex and embedded in human systems, the 'answer' was often prefaced with 'it depends'. This juxtaposition of tradeoffs and probabilistic outcomes was the 'spark' that has brought me here.

How did you get started in your career?

I started my career as an aerospace engineer with a focus in flight control. This led to work in adaptive control systems and eventually autonomy and hybrid inner/outer-loop control - specifically in the area of generating probabilistic experiments and using formal methods to prove airworthiness.

What challenges do you face at Lockheed Martin when it comes to using analytics?

Sometimes the biggest challenge is demystifying what analytics is, and what it isn’t. There’s also a culture shift required - moving from experience and knee-jerk reactions to immersion and exploration of rich insights and situational awareness.

Do you think data scientists can move from industry to industry or do they require a certain amount of expert knowledge?

It depends on the data scientist. If they have a passion for coding, deep learning, and technology itself, then the industry doesn’t matter. I often refer to this as 'back office' data science. However, the greater challenge is the 'front office' data science work. It is this data scientist that is the translator - speaking both the language of the business and the language of data science. The front-office data scientist must know the industry, have a firm grasp of economics and finance, and be able to validate, integrate and use advanced models within a broader decision support framework.

How important do you think it is to build a data driven culture? If so, what advice would you give to companies trying to build one?

We are at a point where data-driven decisions may still offer companies a competitive advantage, however we are likely 3-5 years out from advanced analytics being table stakes and critical to the viability of a company to even remain in business. That being said, if you want to foster data-driven decisions, it needs to start at the highest levels, where employees see that executives won’t commit to a position without the analysis of alternatives and the data that drives it. Be warned however - it is dreadfully easy to be fooled by this modern-day Merlin.

Do you believe machine learning will change the role of the data scientist? Can analytics be automated?

There are already great advances in automating many pieces of the analytics chain, and emerging machine learning technologies are showing great promise. We need to remember however, that 'learning' in this context is but a metaphor. Machines don’t 'learn' the way humans do - as much as we are inspired by biology, the machine is still assessing probabilistic relationships derived from transformations of data. As more of these algorithms become automated, and more purpose-built hardware is created, I believe the role of the (back-office) data scientist will shift from implementation to more integration and more front-office work.

Do you believe companies look at external data as much as they should?

Aside from obvious macroeconomic and specific market data, there are very few external indicators that I’ve found to have real predictive power for the business. The case for retail and ad agencies is obvious as you look at the clickbait we’re bombarded with on a daily basis, but I find social media to be misleading and of questionable value for my particular industry.

How difficult is unstructured data to analyze? Has technology changed this?

Technology has surely made unstructured data easier to analyze - the big question is: 'how useful is the analysis?' There is a lot of art to topic modelling, graph analysis, and even dimensionality reduction and visualization. How many features? Which ones? How many layers in my deep net? How many nodes? What kernel width gives me good separation? What is the relationship among neighbors in 2 nd or 3rd order derived feature space? What in the world did this algorithm just learn? What was my hypothesis again?

How do you feel data science has changed in recent years? What do you see as being the major developments in 2017?

I feel like everyone claims to be a data scientist now, and everyone wants to sell you their shiny analytics tool or automate data science. I think the core density of workstations, the power of GPU’s, and the quality implementation of sophisticated machine learning algorithms has been a boom for doing data science. I’m excited to see what NVidia and AMD bring to the table in 2017.

What will you be discussing in your presentation? Who else are you looking forward to seeing at the summit?

I will present a 10-step recipe for delivering tangible value with data science and advanced analytics. I believe this technique is applicable to any domain and is ultimately the core of any successful analytics endeavor. I look forward to networking with my peers and all of the industry leaders at the event. 

You can hear more from industry leading experts like Walter at the Business Analytics Summit. View the full agenda here.


comments powered byDisqus
Empowering small

Read next:

Empowering The Data Scientist Through Self-Service Data Provisioning