Data science is a relatively new entry into the scientific fold. In general, all of science is concerned with data in the grander scheme of things, but data science has a much broader focus and tends to make much greater use of machine learning and sophisticated algorithms in order to parse data and help people understand it. The field has found great usefulness in the business sector, helping executives understand, in terms both broad and specific, the risks and consequences of important decisions they are required to make for their company.
The field of data science is exploding as companies find its applications indispensable while innumerable job-seekers look to start or restart their careers in the emerging discipline. However, as the supply of eager potential engineers grows, the demand will inevitably become more selective. Recent news has shown that it is simply not enough for a data scientist to be a data scientist any more – everyone in the field, from newcomers to seasoned experts, must begin to look at their responsibilities from a holistic point of view, understanding the problems from top to bottom, in order to truly do their best work.
Why is this the case? Even in other fields of science, basic understanding of the underlying principles that drive the research and data-collection processes is indispensable. A layperson may imagine that hard math-based sciences such as physics and engineering simply involve plugging numbers into formulas and having a computer do the rest of the work, but this is a common misconception that ignores the most important part of science: A firm understanding of the underlying principles of the problem at hand. Given that data scientists are first and foremost scientists, it is understandable that it is not immediately obvious why this could begin to be a problem in the first place.
The main issue is precisely that data scientists see themselves as scientists, not business people. Industry insiders have argued that a lack of understanding of fundamental business principles and inability or unwillingness to put in the social and political work of explaining the hard science concepts to managers and other office workers is a key reason behind the all-too-common failures of data science-driven projects to make a significant positive impact on a given company's bottom line.
There is a bevy of recent commentary and industry news supporting the perspective that data scientists should be given more visibility and responsibility in a corporate setting so that they have the opportunity to explain their work to key decision-makers at the company, as well as more clear expectations that they should understand the business-based fundamentals that underlie their work in the first place to ensure that they can communicate effectively with key decision-makers. It is surprisingly easy to draw false conclusions from otherwise convincing data: That ice cream causes murder, for instance, or that organic food causes autism. Anyone who has an intimate relationship with reality or common sense should immediately see that the idea of causation between these trends is ridiculous, but when it comes to data science in a business setting, any scientist without an understanding of business principles could be unable to make a similar judgment.
Recent events in the business world support the claim that a holistic approach to data science is necessary. A merger between the companies Sisense and Periscope with a combined value in the hundreds of millions of dollars occurred because their founders agreed on the principle that data science should be approached from both ends – the former works from the business side, the latter from the data science side. Even discounting opinions and op-eds from experts within the industry, big moves like this speak volumes louder than words, showing that the executives in charge of the industry's leading companies are more than willing to put their money where their mouth is when it comes to marrying data science to business analytics.
The industry is changing; executives and employees alike will have to change with it. There are certain things that are already obvious or second nature, like using Python or R for data science, or the need for sufficiently developed algorithms that can interface competently with the real world. Scientists – even data scientists in the business world – have frequently been sequestered, doing their work in relative isolation from the other sectors and industries, even employees, that benefit from the results of their research. This often results from a mistaken belief on the part of managers and executives that they need space and a lack of disturbances in order to function effectively, or even that they lack the disposition or attitude to fit in or socialize in the potentially more laid-back office environment. But more and more industry insiders and experts are coming forward and singing the praises of more closely integrating data scientists with the decision-makers at their companies – the reality is that everyone will have to adjust soon.