Business risks of making purely data-driven decisions
Data on its own or out of context with bias and errors can be more bad than good. This is word from Dr Barry Devlin, founder and principal of 9sight Consulting, who is gave the keynote address at the ITWeb Business Intelligence & Analytics Summit 2018, currently underway in Johannesburg.
According to Devlin, although data usually consists of raw facts, often simple numbers or words - it lacks context and often needs information to make sense, "The problem with data is that it lacks context and meaning. If you are going to take data from machines and sensors - which is of course where we are going with the Internet of things - one needs to make allowance for systematic and random errors. Companies now need to worry about calibration, drift and increasing opaque second hand sourcing. These are all serious issues that happen in data collection practice."
He adds that another real danger of becoming fully data-driven is that inaccuracy of the information. "We [data scientists] reach a point where we are also just gathering data and finding information that is interesting to other parties and correlate it. We start looking for correlations that may or may not exist. We need to ask ourselves if this is really data science."
With data being re-used, manipulated and altered - it is dangerous to make purely data-driven decisions as an organisation, says Devlin, adding that business should have a greater understanding of what can go wrong if context is missed or ignored as well as misused and abused. "These decisions which are driven by data only lead businesses and organisations to miss the context that information adds - which can be detrimental."
Distinguishing between data and information, as well as knowledge and meaning, will be a key skill in avoiding the dominance of the data rich, he noted.
"Data-driven decision-making is lazy talk. There is huge value in information - basically data is a subset of information. What we need is information-informed evaluations that allow decision-makers to explore the meaning of the underlying facts and beyond that to include other relevant considerations," concluded Devlin.