The magic of data science

By Joel Nicholson, Marketsoft

The challenges…

  • Does the name conjure images of either laboratory geniuses sitting in top secret Russian military bases or a scene out of the Minority Report of futuristic technicians swiping three dimensional visualisations of data and charts?  What ever the image you have, the definition of data science is wide and varied.
  • Do you know what the real impact of data science is having on your organisation? We hear about the cliched examples such as Amazon’s data science team developing machine learning to improve exponentially on how to predict customer needs and next actionsminority-report
  • Are there any practical approaches for the medium size organisation with limited resources and are at the start of harnessing data science?

The concepts…

  • The cost of data science is coming down. Expertise previously required to model data is now more and more embedded within low-cost software.  For example, regression analysis previously required a university qualified statistician, however is now automatically performed with 1 click.
  • Data quality is still a big barrier for many data science projects. As data is pulled from a broad range of external and internal sources, the gaps in incomplete or unreliable data is only getting larger.
  • Preparing the data for data science activities can consume 80 to 90% of the time for an expensive data scientist, so with limited resources, how can you come prepared?

Actions to consider…

  • Ensure you still have expertise in your team or partners that are strong on creative thinking, yet at the same time relate to business context. The need for pure technicians can now be replaced by low cost software.
  • Think of data science a little like a forensic investigation, there is rarely a single source of golden data on your customers, it is usually about compiling a range of subtle signals that combine together to form a better picture. For example, combine company website statistics on tag management with ANZSIC industry classifications to determine customers that are not keeping up with their industry competitors online
  • Finally, be prepared. Continually deploy data maintenance services to enhance the data, fill the gaps, and make data science as conjured up by the Minority report become BAU!