Prescriptive Analytics: Today’s Fringe of Data Science
Data science today seems to focus a lot on predictive analytics, one of the key differentiators of the field. However, as the demand for insight grows, the expectations of data science’s deliverables grow too. Prescriptive analytics appears to be what most organizations need today and expect to some extent from data science, even if this expectation remains tacit and not clearly communicated. Understanding this aspect of data science and having it in mind when figuring out a solution, be it an in-house one, or a contract-based one, is bound to benefit an organization tangibly, while also provide a more holistic view of data science as a tool for deriving actionable insights that can positively impact your organization’s ROI.
The insights that data science offers today may or may not be sufficient for an organization to benefit from and transform them into a positive impact on its ROI. This doesn’t mean that there is something wrong with the data science process or the value it adds to the data that data scientists handle. However, due to the unreasonably high expectations of data science as a tool / service, and the lack of education of many business people about what it can and cannot deliver, there is a subtle disconnect between the two spheres: the data science world, and the business world. This disconnect, whenever push comes to shove, generates issues that create unnecessary tension between the management and the data science team, that may disrupt negatively the communication and collaboration facets of a data-driven organization.
The problems that stem from this dissonance of data science deliverables and their impact on the bottom line are varied and oftentimes depend on the industry. However, there are some general trends that are observed across various business sectors. For example, the insights the data scientists provide may be interesting and even hint towards certain actions, but they are not solid enough to incite a change of strategy. The latter is often accompanied by a certain risk, which more often than not is an unappealing factor in the business decision that takes into account these insights. So, although educational and perhaps even instrumental in understanding the clients / market of the organization, these insights remain dormant, while the data scientists are asked to solve some other problem, or come up with insights on a different aspect of the company’s data streams.
Implications of these Problems
These problems may not be catastrophic but they create unnecessary tension, while they are conducive to a great deal of unharnessed potential revenue. From the data scientist’s perspective, it may appear that his work is not that important, since the finding he struggles to bring about are not utilized in a meaningful way. Therefore, job satisfaction drops, with whatever unwanted side-effects this entails. From the business’s standpoint, the fact that the insights the data science team provides remain in limbo, translates into missed profit and raises questions like “what do we need so many data scientists for?” and “how else can we justify the presence of this expensive resource?” Clearly, predicting something with the data at hand does not always deliver what’s expected, or if it does, it is not recognized as such.
Prescriptive Analytics is the next level of data analytics that finds today’s data science field as an ideal place to manifest. Basically, it entails not just predicting a certain KPI or some other variable of value to the organization, but also examining how this prediction can impact profit margins or help reduce costs. Going beyond conventional data analysis focusing on the data and questions that can be asked to it, prescriptive analytics asks “what if?” questions that are closer to the understanding of the stakeholders of the project, even if the latter are not trained in this field. Not every data scientist can perform prescriptive analytics, but it is something that every professional in the field can undertake, given enough domain knowledge or expertise, qualities that may not always be discernible from that person’s CV. This is because prescriptive analytics has a lot to do with one’s mindset and communication skills, as well as more subtle qualities, like out-of-the-box thinking.
Even if finding prescriptive analytics in a data science professional seems like finding a unicorn data scientist, it is far easier than that. Yet, in order to pinpoint those professionals who have that approach to data science, it takes much more than counting buzzwords in a CV, or asking candidates about data modeling scenarios. A solid understanding of prescriptive analytics and how it manifests in a data science professional is required from the people evaluating the candidates. Data Science Partnership is one of those places where this kind of expertise is present. Our staff has a lot of experience dealing with all kinds of data science professionals and is able to find those who are capable of offering not just interesting predictions based on your data, but also courses of action that your business developers can take, without the need of a liaison to aid the communication between the two groups. Contact us for more information on how we can help you make prescriptive analytics part of your data science pipeline.