Getting to grips with data
What is a textbook definition of Big Data and Data Analytics?
Prof. Barry Dwolatzky, Wits University's Joburg Centre for Software Engineering: Being professorial, I can say the concept of big data is probably more a buzzword than a thing. The definition has become something that does not look at the size of the data, but, rather, at the bringing together of different types of data and making sense of it. It's really how we joint different sets of data and deal with it. It also has quite a lot to do with how we are collecting data in the digital age. With the explosion of data, the challenge for companies is extracting meaningful information from this huge amount of data from a variety of sources.
Yolanda Smit, regional (Gauteng) director, PBT Group: According to Cindi Howson, the writer of Successful Business Intelligence (BI), big data is the new BI. This is something I fundamentally disagree with. Big data is a type of data. We are still doing analytics on it. We are also still doing BI. Big data and BI are different technologies and the way we deal with that data is different. But, ultimately, the objective is the same, and that is to give insight.
Julie Tregurtha, head of database & data management sales for Africa at SAP: I think it is something new. We always had data. We have always had data platforms and companies that were getting insights from data, but this was in a smaller context. Traditional platforms were never able to cater for the volumes, velocity and variety we are now seeing. The technology has had to move on and we have had to look for new, non-traditional ways to cope with the amount of data being generated. And companies are realising that data is an asset and that it needs to be taken advantage of.
The real problem for business is that they are data-rich and information-poor...Arthur Britz, HTSA
Chris Wiggett, director of Insights and Analytics, Britehouse advanced analytics division: It has always existed. All of these fancy algorithms have been in academia for quite some time. What has happened is business problems have become more complex and more data-driven. We've seen computational powers increase, and along with that, we've seen powerful software bridge the gap between academia and business. But from a semantics perspective, I have a big problem. There are so many terms that people use in reference to what we do - from predictive analytics, to big data, to machine learning, to AI - and, depending on who you speak to, they will use these words. They are essentially all small pieces of a bigger thing. What we've done is called it advanced analytics, which covers all of them. We've been doing this for about two years and we have not seen big data yet.
Brandon Shaban, cloud analytics specialist at Oracle Africa: I think it goes beyond how you collect your data and where it's stored. There's so much data coming in from different areas that employees spend a lot of time consolidating the data. From what we've seen, 80% of the time is spent consolidating data, and the remaining 20% is spent on publishing that data. This ratio should be inversed.
What difference does getting a handle on your data make?
Arthur Britz, CEO of HTSA Pty Ltd, an EOH company: Companies, societies and humans are built for data. It's all around us. But with all this data, the real problem for business is that they are data-rich and information-poor, and we're even poorer when it comes to insight, and we don't know how to handle that.
Jeremy Potgieter, regional director of Eseye: I agree with you. The data in itself is not information. It's how you take that data stream and identify it against a defined strategy. The strategy for me is what's important. How does a company identify this strategy? What is it that they want to extract out of this data? And how does this single stream of data once analysed speak towards multiple facets inside your company? There is a lot that needs to be understood. But most companies that go into a data analytics strategy aren't necessarily looking at the right data. So the first step is to define the strategy and get the right people to understand the data before you make a decision on what that insight is you are looking for.
Gary Allemann, MD at Master Data Management: I like to use the term 'time to insight'. That's what big data brings us. Historically, BI has been quite a long process, and often, by the time you get your answer, it's no longer relevant.
There's so much data coming in from different areas that employees spend too much time consolidating the data.Brandon Shaban, Oracle Africa
This is where big data is different. It consolidates a variety of data quickly, enabling you to ask and answer questions quickly. This is where big data is different, as it enables you to take on problems traditional BI could not address.
Gustav Piater, marketing & sales director, AIGS (Yellowfin BI South Africa): For me, the biggest change is the audience. Traditionally, big data used to be the analysts, but now it's the general businesses person who needs access to more sets of information.
What's driving the growth in big data and data analytics?
Lieb Liebenberg, CEO at ITSI: Computational power has made a lot possible. I work with learners and it was always a mystery to us as to what happens when they study at home. We have been able to figure this out as we have set up 'competitive neural networks' for the learners, which generated data and, in turn, helped them figure out better ways to study. This was not possible a few years ago. Even though we don't have a lot of data, the computational power has made it possible.
With all this data, how do you avoid information overload. Is having AI take over from people a solution?
Dr Caroline Belrose, Accenture's chief data scientist and MD for Accenture Analytics: You don't take people out of the equation, but some of the big questions include, 'when do you give stuff to people and when do you give stuff to machines?' AI can be very helpful in managing information. For instance, I would really love to have something that would help me manage my email.
What do you want from big data and data analytics?
Lieb Liebenberg, ITSI: For me, I want predictions. I don't care about the past. I want to know, 'given what has happened so far, this is what is going to happen'. Already we can see some patterns. One of the local universities noticed that the biggest indicator of whether someone is going to pass or fail is whether they log in to see the results of a particular test. It found that 70% of those who log in actually pass. But we want more data than that. We want to be more predictive.
Arthur Britz, HTSA: The humans also have to come to the table. We have to increase the quality of our thinking. This means we have to ask better questions. Right now, we are hoarding data just in case there is some significance we don't understand. By not improving the quality of our thinking, or asking better questions and just hoarding data, I think the return on our investment will be very low.
Chris Wiggett, Britehouse: I absolutely agree. We are trying to answer new questions with old metrics. And from an insights perspective, we must be brave enough to push the envelope. You have to find new ways of measuring new questions. I thinking if you push that, then perhaps you will realise that all data is noise - when you start out. This will prevent you from assuming certain data has more value than others, and also from making wrong assumptions. With all the big data and data analytics tools now available, you can start by assuming everything is noise, and so not make any mistakes related to either institutional memory or knowledge.
If you are not asking the right questions, you are not going to get anything useful.Dr Caroline Belrose, Accenture
Dr Caroline Belrose, Accenture Analytics: If you are not asking the right questions, you are not going to get anything useful. I see companies investing in analytics thinking, 'Oh, you're going to give me the answers to solving my business'. No. Having great analytics is not a solution to having no strategy. If it does not solve a business need, it's just a waste of time.
What do you have to be careful about when it comes to managing data?
Gary Allemann, Master Data Management: Part of the problem in managing the complexity is managing the source. Is our data good enough? Where is it coming from? Will the right people have access to it? I'm seeing a lot of investment from our clients in data governance, so they can get real value out of big data and advanced analytics. So if you are going into this sort of journey, without getting the fundamentals in place, then you are really wasting your time.
How will big data impact our lives?
Prof. Barry Dwolatzky, Wits University: There is something called 'technological determinism,' which is the belief that technology changes the world. My belief is the reverse. Is the world prepared to be changed by technology, and is this the right time? If a computer had been invented in the 1800s, it probably would not have any impact at all. The state of society then was not at a point when a computer would have made sense. I think it's the same with machine learning and big data. Is the world at a stage where it need this kind of stuff? I think we are because of the complexity of our society. It will make a huge difference because people at every level of society are struggling to cope with the amount of information in their lives, and the technology around big data helps them to deal with all this information. It will make a difference, but I wouldn't hazard a guess if it will be positive or negative.
Ryan Naude, manager of the data solutions division at Entelect Software: I want to back up the Prof.'s point. Take for example the first person who bought a fax machine. It was probably useless to them because there was no one to send a fax to. But when it comes to big data, I do think it's making a difference in our lives already. Just look at traffic apps. They will alert you about accidents. This is already making a massive difference in my life as it prevents me from being late for a meeting.
This article was first published in the November 2017 edition of ITWeb Brainstorm magazine. To read more, go to the Brainstorm website.