Getting the board’s approval for a big data project is a good start for CIOs but there are four challenges which must be addressed first, according to a data expert.
Speaking at the IDC Asia-Pacific Business Analytics conference in Sydney recently, NICTA senior researcher Dr Rami Mukhtar shared his top four big data tips with delegates.
Extracting true value
According to Mukhtar, one of the biggest misunderstandings with big data is that people perceive it as a “crystal ball” which will tell them what is going on in the business.
“That couldn’t be further from the truth. True value from big data can only be extracted when you have a very precise business problem which you are trying to address,” he said.
The business problem needs to be understood and well defined first, he said.
“Executives need to ask, 'How much revenue will I gain from applying the solution to this business problem?'”
Business to data mapping
Mukhtar told delegates that solving a business problem using big data has to work within the context of a business process.
“I may understand that my particular solution within this process is going to be better business predictions,” he said.
“If you can’t make the business process match with your data assets than I would suggest to you that this is going to be a very difficult problem to solve -- even with big data.”
Analysing big data
The third challenge is getting the company’s data scientist to examine the data and extract critical information such as customer buying habits.
“Big data is not like a data warehouse, there are no well-structured tables with well-defined relationships,” Mukhtar said.
“All we have in a big data solution is a set of flat files.”
To overcome this, he suggested enterprises use catalogue data assets which will make it easier for data scientists to “cheery pick” critical data.
“For example, if we know what 10 per cent of the [data] file means and this percentage is relevant to the business process that we are addressing, this is absolutely fine,” he said.
“Corporations have privacy policies where they tell the customer exactly what they are going to use that data for,” he said.
“The ways we have seen privacy concerns addressed is to have extremely tight control on which data scientist is looking at the data or using technology to blank out sensitive data such as customer names,” he said.
Follow Hamish Barwick on Twitter: @HamishBarwick