Learning’s as a Data Scientist

They say science is like magic, but real… then the magicians of the 21st century who dazzle and delight are the data scientists! I have been working as a data scientist for over three years now and would like to share some key principles that I have learned and inculcated into my thinking which has proven to improve my delivery as I continue to grow.

1.      Solving the problem at its root cause

Often as a data scientist you are asked to solve a problem that business has been experiencing for a long time. You are expected to make the magic happen. [Yes magic, even though we both know that it’s just a bunch of clever math that creates the illusion of magic].

In the opening act of your project, you have the best opportunity to actually figure out what the potential root causes of the problem are [because you’re still waiting for data or a volunteer to saw in half]. Use this time to do two things really well: 

·        Understand the business as a whole, which will give you better context as to how solving this one problem relates to supporting the business from a strategic perspective;

·        Understand what areas your problem affects in the business and what the potential implications of a deployment could mean. The model you are building could have no impact on any process [which should cause you to question why you’re doing what you’re doing] or it has a large impact and therefore you need to start preparing the business for potential change.

A word of caution…Nailing these two tasks really sets you up for success, but be careful not to set unrealistic expectations through these exercises so that you don’t end up underdelivering the magic. No one wants to see a mouse pulled out of a hat when they’re expecting a rabbit!

2.      Use the principle of continuous delivery to your advantage

As you go through the process of preparing the data for modelling and feature engineering, I have learned that playing this back to business in the form of frequent insights really helps the process [both for business and for you]. Let me explain, when you produce insights from the data you are either going to:

·        Confirm what business already “knows” which means you will delight them by confirming their gut feel and confirm that you’re on the right track, OR;

·        Show them something that contradicts what they “know”. This way you can confirm that you’re working with the right data and start showcasing some value to business by challenging their gut feel. [Good luck, some of these sessions are quite tough but ultimately lead to a better solution]

In both of these cases, however, the real benefit is receiving and including stakeholder feedback into your solution FAST. This helps guide your thinking and validate most of what you’re doing up front, showing continuous value.  Setting your audience up for the big reveal.

3.      Use the latest and greatest tech

In my career, I have been fortunate enough to get to play around with and use the latest and greatest in technology to assist in solving problems. This has really helped me gain a better understanding of where the data science field is heading and my potential role in the future. Here’s the caveat, however: all businesses are not necessarily ready for what you are able to deliver to them. In order for your project to be successful, you will need to assess what the business can handle and what they actually need at the moment [remember that point upfront about understanding the business - it comes in handy here]. In some scenarios, your solution may be better off if it has a smaller impact on business as a start which then evolves into a larger solution. This allows for your solution to first prove itself. Once it has been proved, there will be little resistance to change as it scales up and delivers more and more value.

In conclusion ensure that you know what you are solving for, use your time wisely taking your stakeholders on the journey with you and align what you are doing to the bigger picture…[I could have just written those three points, but where’s the magic in that?]