Data science and Business Intelligence

After the hype around big data a few years ago, a new term is popping up more and more in business: data science. Almost everyone will come into contact with it in their daily life, for example when using web shops. Frank de Nijs is business developer at Centric; he explains what the differences are and what the concepts have in common.

Business Intelligence: insight into the results

"Business Intelligence (BI) is based on looking back," he says. "By looking back you can draw conclusions with statistics from the past"’ You can see certain trends taking place through BI, for example an annual increase of 2%. "If that stays the same for years or goes through a steady change, you can conclude that the increase will also be 2% the next year or will deviate from it at a comparable pace.".’

Data science examines the background of the trends

With BI you visualize trends by looking back. But data scientists go further, and as a first step try to understand the causes behind those trends. De Nijs: "You look for the characteristics behind such a trend: which characteristics of the company ensure that annual increase of 2%, or what causes that fluctuation? That way you can more accurately predict what will happen in the future, if you are also able to monitor those influencing factors."

Pros and cons of data science compared to Business Intelligence

"Moreover, data science works with enormous amounts of data, which are analyzed with the help of algorithms and Machine Learning. Those methods are able to work their way through huge amounts of data, discovering connections in it.

That is a major advantage of data science compared to BI,' explains De Nijs. The characteristic of BI is that the analyst has to decide for himself for which data the causality has to be proven. Data science can itself find factors from an incredibly large amount of possibilities that you never thought would have any influence. Think of tensions in a neighborhood, about which you receive signals as a municipality. You wonder: what sources are available about this? Healthcare costs, reports to Youth Care, things that happen with the police. These can all be factors that can explain certain trends. You will organize that data by using data science algorithms to discover which characteristics are dominant in which circumstances."

Working reactively with BI versus working proactively with Data Science

Working with BI, which allows you to look at the past and make decisions based on that, is in principle reactive. One of the great advantages of Data Science compared to BI is that you can implement proactive policy.

De Nijs: "A data scientist can receive live information and intervene when predictions show that an operation is going in the wrong direction. You can also predict the consequences of that intervention using models and algorithms. This way you don't wait, as with BI, for the results of the past month to come in, but you can already make an adjustment in a way of acting in the workplace halfway through the month."

Data science is not always preferred over BI

Data science's algorithms can find their way through huge amounts of information. But when less data is available, the BI analysts have the advantage again. "Then the data science models cannot give any significance", says De Nijs. “In those cases, a single outlier can lead to a completely different model. Then you still need the human insight of a BI expert, who uses his knowledge to determine whether a relationship is causal or not."

Always in touch with Data Science

Almost everyone will come into contact with data science in their daily life - desired or unwanted. "When you buy something via the internet, a web shop analyzes a lot", explains De Nijs. "Go to certain web shops on an expensive device, then the prices are higher than if you do that on a cheaper laptop or smartphone."

Price elasticity is also data science. "It also happens in the travel industry: when you look at the same hotel room for the second time, it can suddenly be a bit more expensive."

Synergy between data science and BI

The rise of data science does not have to mean the end of BI, De Nijs expects. In fact, both techniques work best when used in combination.

“Organizations must first be able to determine exactly what is happening before you can investigate the possible cause of a problem. And that is real BI work,” he says. “An analyst tries to see if there are rising costs, decreased productivity, that sort of thing. Then you go with a Data Scientist to see if you can find out the causes of those problems and come up with advice on how to intervene proactively. But at the end of the month you always want to know whether those recommendations have actually led to improvement, and you can do that via BI."