Data Science in 2017 Will Be All About Providing Value

[Please note that many many more items could be added to the list below... let's start adding more in the comments!!]

2016 was a big year for the data science field. Data scientists had the best job in the country, and market research company Forrester predicted that businesses embracing data science today will see a return on that investment to the tune of $1.2 trillion in revenue in just a few years, effectively propelling the field from buzzworthy topic to the forefront of business strategy. Last December, we predicted 2016 would be a banner year for data science, with major developments in deep learning, web integration, education, and overall reach of the field — much of which came to pass.

At a glance, it seems that 2016 will be a tough act to follow for the industry as a whole. But while this year was all about validating the importance of data science and predictive analytics, 2017 is poised to bring about a new focus on the true value of data science on an enterprise level. Below are a few trends we expect to see in 2017 that reflect that change.

Data science platforms will become a must have for enterprise companies looking to scale data science operations.

Platforms designed to identify insights in big data and leverage those insights across all aspects of a business were 2016’s top emerging technology. In 2017, platform adoption is likely going to reach a fever pitch as enterprise companies attempt to scale data science efforts and move toward decision making that relies on predictive modeling rather than gut instinct.

Even for companies already performing data science at a high volume, platforms are helping data science teams overcome two common barriers: a disjointed, inefficient workflow and a lack of institutionalized knowledge. Teams that work together in a shared space achieve better collaboration and reproducibility, reducing unnecessary duplicate work and producing results faster. And with code libraries, visualizations, and data models available in one place, key contributions won’t go unnoticed or unused. [Shameless plug for the Data Science Enterprise Platform - DataScience Cloud.]

More businesses will launch bots, and the resulting data will become a data science gold mine.

2017 will see bots passing the Turing test to fool us into thinking they’re human. Companies have invested billions to create bots that learn about consumer behavior through natural language processing and to analyze the resulting data, which is now being generated by millions of users around the globe. Far from being a specialized subset, bot data contains all of the hallmarks of user behavior data that’s been collected since the dawn of the Internet, and that data is a gold mine for data scientists.

Bot data constitutes a new, rich data stream that is inherently social and optimized for feeding machine learning algorithms. Data scientists can use bots on a platform like Facebook Messenger to capture user demographics, traffic rates, sales conversions, or any activity or API request, and leverage that data to identify promising market segments or track traffic and sales patterns.

CMOs will effectively become data scientists.

Do chief marketing officers need to be data scientists? The question has been asked before, but never has there been a time when effective marketing was so deeply tied to technology. To stay ahead, CMOs now need to leverage tech and data to get the right message in front of the right customer at the right time — and measure the outcome of those efforts accurately.

CMOs are already regularly using data science techniques to make campaigns more effective. Customer lifetime value modeling can segment customers by their behavior — not just by their demographics — giving marketing executives the power to send highly targeted messaging. Churn models can help identify customers who might unsubscribe or stop shopping, which can inform retention strategies. We believe that in 2017, the role of marketing executive will become synonymous with utilizing customer data to its fullest extent.

Probabilistic languages and tools will gain popularity, allowing data scientists to tackle more complex models in less time.  

Probabilistic programming aims to do in a few lines of code what takes other languages thousands, which is why the Defense Advanced Research Projects Agency (DARPA) made it the focus of a four-year project to improve machine learning for the masses. And even though the project ends in 2017, probabilistic programming will continue to gain popularity next year as tools become more feature complete and language APIs improve.

Model building is one core deliverable of what data scientists do, and if that process is more intuitive, data science work will accelerate. Probabilistic programming languages (PPLs) automate much of the computational work associated with probabilistic models and machine learning, allowing data scientists to focus their efforts on formulating mathematical problems. And although probabilistic models often require large amounts of memory to run, a growing number of cloud-based solutions are helping data scientists overcome that roadblock in 2017.

The number of smart cities will continue to grow, and the need for data science in government will grow with it.

Cities are increasingly using Internet of Things (IoT) solutions and other communication technology to tackle a number of problems, from traffic congestion to climate change to crime. These so-called smart cities are making that data available publicly to promote innovation, thanks in part to $160 million dollars in federal funding offered through the White House’s Smart Cities Initiative this year.

The goal of these projects are to encourage engineers, scientists, and policymakers to work together in order to improve life in urban areas. The driving force behind this innovation is data science, which will continue to become more important in urban planning as data collection — from nodes on traffic signals and even sensors on U.S. Postal Service trucks — collect real-time data on environment, infrastructure, and activity.

There are dozens of other data science trends we could point to, but the one overarching theme is this: Most companies and government agencies have already come to see the importance of big data and data science. In 2017, improvements in the tools and technology we use to perform data science will continue to refine the field — and help practitioners provide real, concrete value.