A good data analyst should embody three roles – that of the developer, statistician, and domain expert.
Within the data science community, Drew Conway’s data science Venn diagram is popular. I provide my interpretation below.
Especially given the trends towards Big Data characterised by its 3Vs (volume, variety, velocity), coding skills are increasingly important. Without a basic grasp of coding skills, it would be very difficult to collect, cleanse, manipulate, and analyse the data collected.
Without coding skills, a data analyst would face issues with data quality and quantity, and would be operating in the space of a traditional researcher (think “professor”).
We already think of a data analyst as a developer with programming skills. And my experience is that many clients or hiring managers focus too much on this skill. But other hard skills are also needed. Read on…
A fundamental understanding of mathematics and statistics is needed to make sense of inputs, outputs, and to draw useful and correct conclusions. Data analytics is the science (or art) of extracting useful information from data – and statistics is vital to this process.
A lack of statistical knowledge may mean that incorrect conclusions are drawn as the correct conclusion may sometimes be counter-intuitive.
A good data analyst operating in customer research needs to have expertise (or at least the flexibility to operate) in that sales and marketing environment. One needs to understand what experiments are being conducted (e.g. A/B testing), any practical limitations, and also relevant business concepts such as customer lifetime value, churn, market segmentation, etc.
This allows data analysis strategies to be planned, and for insights to be directly applicable and actionable.
A good data analyst should have the broad mix of skills needed to be able to fit comfortably into those three roles – Developer, Statistician, and Domain Expert. Such a data analyst is best placed to deliver robust, business-specific, and actionable insights for his clients.