Connecting...

W1siziisimnvbxbpbgvkx3rozw1lx2fzc2v0cy9zawduawz5lxrly2hub2xvz3kvanbnl2jhbm5lci1kzwzhdwx0lmpwzyjdxq

Beyond the Numbers: The Business Side of Data Science by Shawn Simpson

W1siziisijiwmtgvmdgvmjavmtuvndyvmjuvnju5l3rhcgfkigrhdgugc2npzw5jzs5wbmcixsxbinailcj0ahvtyiisijkwmhg5mdbcdtawm2uixv0

Image credit tapad engineering

If you work in Data Science you can most likely relate to this article. Shawn Simpson tells us about her time working in the Tapad Data Science team and what she has learnt on how the better you understand how your work impacts the business and vice versa, the more doors will open for you.


'It’s been a little over a year since I joined Tapad’s data science team, and I’ve learned a lot over that time. I was originally drawn to Tapad because of their highly skilled team, collaborative environment, and modern tech stack — including Big Data tools like Spark and Hadoop.

Since then, the company has grown and changed in ways that have motivated us to adapt and improve. This includes big moments, such as the Tapad acquisition in early 2016, which has opened the door for new and exciting projects. It also includes other changes that have led to interesting tasks, such as finding the best tools for our data science work.

Throughout all of this, I’ve been noticing a trend that I think is important for any data scientist to know — the better you understand how your work impacts the business and vice versa, the more doors will open for you. I’m learning more about this everyday, but here are a few thoughts that might get find new opportunities within your own company:

  1. Communication is key. Data scientists interact with a wide range of people — from engineers and product managers to sales and end client users. At Tapad, the data science team has found itself more engrained in business efforts and talking more with end clients as a result. To be effective in this role, it is crucial that you can convey the key concepts of your work using primary terms and ideas. When you build a predictive model, what’s really going on when you cut away the jargon and the math? My past experience as a Columbia professor has been enormously helpful here. When you teach something, you have to distill it down to its essence. Student questions force you view problems from many perspectives, and to find quick, explanatory examples. If your company offers speaker training or ways to help prep, take them up on it. Practice teaching peers about what data scientists do. You’ll find doors open up if you are able to speak with colleagues and clients in a clear, accessible way.
  2. Don’t just analyze, consult. A good data scientist doesn’t just perform data analysis in a vacuum. A key part of the job is translating business or end client problems into actual data needs. This offers the opportunity to be a consultant to your company’s clients and business leaders. If this translation isn’t done in a thoughtful way, it is easy to end up building the wrong models with the wrong data. Be upfront with your questions and realize that, as a data scientist, it is your job to stand up for what your data can or can’t do. What is the exact problem we are trying to solve? What signal should this translate to in terms of data? Are there potential biases here? What filters or time windows need to be applied? What are the consequences of errors? Be persistent in consulting clients until you feel you have a full picture of what needs to be done from a data perspective. Not only will this build up your value as a data scientist, but as a consultant to the business as well.

  3. Evolve your tech stack with the business. Engineers and data scientists at Tapad code predominantly in Scala, and our data scientists have traditionally used Scalding jobs for ETLs and Spark ML/MLlib for modeling. However, over time we have evolved our technologies as business needs have changed. For instance, we are in the process of migrating from an on-prem cluster to using Google Cloud. This opens up new possibilities for our data science stack — such as Google Datalab, Big Query, or Tensorflow — and will allow us to scale ephemeral clusters optimally for the job at hand. Instead of resisting change, we are doing more exploration of tools while staying true to the big data stack that has worked for us. In the end, you’ll find ways to embrace the new without giving up what’s already working.

By prioritizing the relationship between data science and business, I’ve had the opportunity to work on some fascinating projects at Tapad — and even travel abroad to countries like Thailand to talk with end clients and other data scientists! If you find yourself in a rut or uninspired in your current job, look around at ways you could be in a more consultative position or try to hone new skillsets — such as public speaking and communication — and take your data science expertise to the next level.'

This article was written by Shawn Simpson and posted originally on engineering.tapad.com