Data analyst vs scientist - where does one end and the other begin?

21 Sep 2016

  • Analytics
  • Data

By: Andrew Birmingham for ADMA

With regard to how a business analyst might use visualisation compared to a data scientist, the biggest difference you would see is not technical skill but the thought process. Business analysts are  typically process oriented thinkers, who look for changes or breaks in a pattern and then record them, but are rarely asked to explain ‘why’. A top notch business analyst is incredibly skilled at teasing out and understanding processes and patterns, and will tend to use data visualisation in a descriptive way, to share what they found with stakeholders. A data scientist will use it in an exploratory way, to tease out potential trends for analysis, or to present a story with an ‘a-ha’ moment. Both are great and valid uses of visualisation, they just reflect different skill sets and thinking styles.

Good business analysts and data scientists are seeking to tell a story with visualisation, and both require broad skills much more than specialist niches. If we think of a business analyst in this context as being a business data analyst or a business intelligence professional then the application of their data skills would usually be focused on ensuring completeness, accuracy, consistency and perhaps even audibility of their data sets and especially (but not necessarily) big data sets. They would then typically produce visualisations which told the story of the data, often in the form of bar charts, pie charts and heat maps.

By contrast data scientists are curious by nature and construct data experiments to understand – in a scientifically robust manner – how human behaviour, environmental patterns or market activity moves, and how certain factors may influence it. According to KINSHIP digital Victoria GM Walter Adamson, “In that regard, a data scientist can be viewed in the same vein as an economist or a clinical researcher – it’s about having an idea, forming a hypothesis and then constructing an experimental design that controls for as many extraneous variables as possible.”

He says data scientists firstly explore the data in terms of structure, statistical distributions and relationships. They initially use visualisation to explore the data. During this exploration phase the business questions would be tested against the data, various shapes, encodings and structures of the data, and various visualisation. It may be that methods other than statistical methods are necessary to gain insights into the data and this is also the realm of the data scientists. The key approach to visualisation for data scientists is to ensure that visualisation doesn’t miss the point of the story – if the plot misses the point then they’ve “missed the plot”. Fundamentally, for data scientists, visualisation combines statistics and design to make meaningful figures. This guides the path from the exploratory phase to the explanatory phase and the explanatory visualisation.

CATEGORY Analytics Data

TYPE Article