Data scientist, according to Glassdoor, will be the number one job in America in 2016, with over 1,700 job openings and a median base salary of $116,840. This is consistent with Indeed.com, whose 2015 first quarter data saw the number of job postings for data scientist grow 57% as compared to the analogous quarter of the previous year. It’s no surprise then, as we’ve mentioned before, that Data Scientist has been deemed the “sexiest job of the 21st century.” But the term is ambiguous at best. Asking ten different employers how they see the role of data scientist may yield ten different responses.
Generally speaking, data scientist tends to encompass broad programming knowledge, statistical analysis and predictive modeling. In particular, predictive analytics are proving to be an especially paramount subset of data science in today’s business environment. The designation of ‘predictive analytics’ is relatively new one, coming to the forefront only within the last decade or so, but has its roots in many notable and long-established fields including statistics, machine learning and data mining. Thus, many of the predictive analytics techniques are not new; but how the fast-growing field of predictive analytics is shaping the future of business is revolutionary. Predictive analytics applied within the business arena is often referred to as business analytics, though the strict definition may depend on who you ask.
Predictive analytics, in its most straightforward form, is simply the process of discovering interesting and meaningful patterns in data that can, in turn, produce actionable insight. This field will be instrumental in guiding future business undertakings, serving to augment traditional business intelligence (BI) contributions.
It is important to note that predictive analytics is in no way divorced from business intelligence efforts, but neither are they synonymous. Business intelligence is a retrospective analysis and, consequently, descriptive in nature. Predictive analytics, on the other hand, involves future analysis and is, as one might predict (excuse the pun) from its moniker, prospective in nature. Outputs of BI analyses are typically reports or dashboards that summarize important, predetermined characteristics of the data through Key Performance Indicators (KPIs). These KPI reports are user-driven, the crucial indicators being determined upfront by analysts and decision makers. Business analytics, on the other hand, is data-driven. Predictive models are able to actually identify which variables will prove predictive and how well they will predict the target.
Where a BI report might answer questions such as “what promotions did a client respond to,” or “which products had the highest click-through rates in the past month,” predictive analytics answers questions like “which promotions is the client most likely to respond to” and “what is the likelihood a client will click-through a particular link?” The differences may seem nuanced, but they’re actually monumental – they may be the margin between success and failure in a business undertaking. Ultimately, predictive analytics drive informed decision-making. While BI may tell you which are your most valuable customers or products today, predictive analytics will tell you which customers and products have the potential to be the most valuable tomorrow.
Whether you are aware of it or not, predictive analytics is already shaping your life. Likely, for the layman this effect is taking place most obviously through targeted marketing. Over $100 million in venture capital has already been invested in predictive marketing companies, underscoring the rise of predictive marketing. More and more businesses are realizing that data-driven decision making, enabled by predictive modeling, can aid in cross-selling and upselling, as well as help to identify and target new customers, and direct existing ones based on revealed habits. The quintessential example of analytics producing potent targeted marketing efforts came when Target famously identified a teen as pregnant, based on shaping habits, even before her dad knew the truth.
Of course, many consumers don’t always appreciate that level of personal exposure, so strategies have been developed to target customers more subtlely, which really highlights how predictive analytics is as much an art as it is a science. It demands a certain level of inquisitiveness, intellectual curiosity, and high-functioning intuition. As the above story indicates, while the predictive model itself was successful, it had to be tempered with human awareness and sensitivity.
Furthermore, being a successful data scientist in the realm of predictive analytics requires a deep understanding of the data. For example, data preparation can take up to 60-90% of the time demanded for the entire modeling process, and because the models are data-driven, if the data is not good neither will be your models or subsequent conclusions. Equally critical is a deep understanding of the business objectives. Because analytics are used to drive informed decision-making, predictive models must ultimately relate back to the business understanding of the project if they are to be useful. Simply investing in predictive analytics will not automatically yield great results.
Data scientists who can exercise business acumen with the technical know-how are in increasingly high demand as the use of predictive analytics no longer belongs to a niche group of large organizations. With the increasing availability of computers with high computing power, coupled with the accessibility of open-source softwares (such as R or Python), it is becoming instrumental to driving decision-making for most medium-to-large organizations, and many small ones as well. Utilized alongside traditional BI reports, it is helping generate improvements in efficiency, decision-making, ROI, and ultimately the bottom line. Gartner has predicted that by this year, “70 percent of the most profitable companies will manage their business processes using real-time predictive analytics or extreme collaboration.” Additionally, a Forbes Insights’ study showed that 86% of predictive analytics users experienced positive return on investment. The proof is in the pudding – data-driven decision making executed aptly is delivering results. Astute universities are acknowledging this burgeoning talent demand and are now offering masters programs in business analytics, just as wise companies are using predictive analytics technologies to augment their existing BI endeavours.
In order for companies to successfully implement predictive analytics measures, it is imperative that they are drawing on clean, high-quality data to build their models. Usable, high-quality data is achieved when there is a solid understanding of the different data dimensions, which is realized through data profiling. This highlights the necessity of developing an effective data strategy. At Apoorva, we offer data strategy solutions in the realms of data governance, data usage, data integration, data consolidation, data privileges, and data storage structure. Having an effective data strategy in place now will save you time and resources down the road, as you go to put your data to good use. Contact us today to see how we can help you get the best out of IT.